State financing of public higher education using productivity values

Material Information

State financing of public higher education using productivity values
Cunningham, Stephanie
Publication Date:
Physical Description:
xi, 189 leaves : illustrations ; 29 cm


Subjects / Keywords:
Education, Higher -- Finance -- Colorado ( lcsh )
Public universities and colleges -- Finance -- Colorado ( lcsh )
Education, Higher -- Finance ( fast )
Public universities and colleges -- Finance ( fast )
Colorado ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references (leaves 168-189).
General Note:
Submitted in partial fulfillment of the requirements for the degree, Doctor of Philosophy, Public Administration.
General Note:
School of Public Affairs
Statement of Responsibility:
by Stephanie Cunningham.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
34533607 ( OCLC )
LD1190.P86 1994d .C66 ( lcc )

Full Text
Stephanie Cunningham
B.S., Colorado State University, 1970
M.B.A., Colorado State University, 1986
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Public Administration

This Thesis for the Doctor of Philosophy
degree by
Stephanie Cunningham
has been approved for the
Graduate School of Public Affairs
!3 Qt&oMi

Cunningham, Stephanie (Ph.D., Public Administration)
State Financing of Public Higher Education using Productivity Values
Thesis directed by Professor Peter deLeon
This thesis utilized human capital theory to develop a method to fund public
higher education institutions based on the increase in former students incomes due
to education. Regression equations consisting of independent variables from
statewide student data bases and a dependent variable of income provided by
unemployment insurance files calculated returns for a variety of public higher
1 education services: e.g., degrees, classes, disciplines. The regression returns were
modified in a cost/benefit matrix to father analyze the return to the states investment
in public higher education. Some adjustments decreased returns for high cost capital
construction requirements, or due to former students migration out of the state, while
returns were increased for serving low income individuals.
The returns to investment estimated in the regression equation are tentative
because of the findings that 1) a longitudinal methodology is needed, and 2) this data
set indicates that at least 1,000 records of each specification must be available before
the regression calculation is likely to generate a statistically significant return on the
cost variables. Although this was an exploratory study, the methodology was found
to be feasible if sufficient numbers of records are available.
This study measured the states return based on dollars allocated to higher
education, not the individuals return based on time in school. Within this context,

the study found preliminarily that the state does receive a small but statistically
significant return on its investment: an increase in income of 2.7% in the first year
after a student leaves school. Different disciplines generated different returns:
business, sciences, engineering, and arts and humanities all generated a return to the
state based on every $1,000 of state investment. Minorities generated larger returns
in the first year after leaving school than Caucasians and Asians. Males generated
larger returns at the baccalaureate level, while females generated larger returns for
certificate, associate degrees, and class-taking.
If a longitudinal methodology were employed and the results were confirmed,
then these results could be used to change the budget distribution system to reflect
the most productive areas for state investment. If preliminary results were confirmed
by additional years of data, baccalaureate and masters degrees would receive the
highest priority for funding and PhDs and class-taking would receive the lowest
The abstract accurately represents the context of the candidates thesis. I recommend
its publication.
Peter deLeon

Many individuals assisted in bringing this venture to fruition. A special thank
you to:
My committee members, for their sharing of expertise and
Peter deLeon, chairperson;
Laura Argys;
Linda deLeon;
Roslyn Elms;
Robert Gage;
Gregory Smith, for his expertise of unemployment files;
Patty OConnor, for her SURDS expertise;
James Podolak, for his statistical expertise;
Richard Gritz, for his census data knowledge,
David Stevens, for his sharing of unemployment file data;
James E. Sulton, Jr., for his help and guidance.
This thesis could not have happened without the unselfish support and caring
of my family Alan, Stacy, and Sara.

1. INTRODUCTION and STATEMENT OF SIGNIFICANCE....................... 1
Purpose ........................................................ 1
Scope of the Study.............................................. 8
Organization of the Thesis .....................................11
2. LITERATURE SURVEY............................................... 13
Public Sector Productivity .....................................14
Productivity Strategies ................................. 17
Public Higher Educations Environment.................... 23
Improving Public Higher Educations Productivity......... 32
Public Budgeting Models.........................................34
History of Public Budgeting...............................34
Government Support for Higher Education ................. 40
Colorados Budgeting Practices............................43
New Budgeting Practices...................................44
Human Capital Theory ...........................................49
Economic Productivity and Human Capital.................. 49

The Individuals Decision to Invest in Human Capital......... 51
Empirical Specification and Results ......................... 65
Empirical Estimates of Returns to Postsecondary Education ... 66
Linking the Individuals Decision to Social Rates of Return ... 71
Related Concepts from All Three Approaches ........................ 77
3. RESEARCH HYPOTHESES AND METHODOLOGY................................ 79
Primary Hypothesis .................................................81
Secondary Hypotheses................................................82
Design .............................................................83
Return to Education Regression Equation............... 84
Costs and Benefits Matrix............................. 93
Internal and External Validity ....................... 98
4. RESULTS AND ANALYSIS ..............................................102
Characteristics of the Variables.............................102
Regression Equation Results..................................110
Regression Equations .................................110
Multicollinearity ....................................110
Results ..............................................Ill

Cost/Benefit Analysis ..............................130
Primary Hypothesis .................................131
Secondary Hypotheses................................134
Conclusion ................................................139
5. CONCLUSION......................................................142
Methodological Considerations .............................144
Human Capital Theory................................144
Financial Aid ......................................149
Capital Construction................................150
Theoretical Considerations.................................151
Policy Implications .......................................157
Political Considerations ...........................161
Institutional Considerations........................165
Implementation Considerations.......................165
Conclusion ................................................166

2.1 Lifetime Income by Education......................... 56
2.2 Income Profiles by Education and Race ............... 58
2.3 Wage Premiumns for Job Entrants ..................... 68
3.1 Revenue Streams Supporting Higher Education.......... 79
4.1 Average Experience Graphed by Years..................109

2.1 Alternative Versions of Productivity............................23
2.2 Types of Budgets ...............................................34
2.3 Returns to Education............................................67
3.1 Raw Data Elements..............................................101
4.1 File Comparison ABL .........................................104
4.2 File Comparison NABL ........................................105
4.3 Variable Means for Each File...................................108
4.4 Unstandardized B Coefficients on All Significant Variables....112
4.5 Unstandardized B Coefficients on Total Files ..................114
4.6 CoveH...................................................... .118
4.7 Gender.........................................................120
4.8 Ethnicity......................................................123
4.9 Degrees........................................................126
4.10 Classes.......................................................128
4.11 Per Cent of Significant Cost Coefficients on Number of Records 129
4.12 Total Return to the State for Investing in Public Higher Education 131
4.13 Discipline Cost Benefit Model.................................136
4.14 Disciplines Returns per $1,000 State Funds................. 136
4.15 Disciplines...................................................140
4.16 Average Costs for Degrees and Classes.........................141

... We dont seem to know what the kids are learning, what they
actually know, and what they can do as a result of going to school
.... If the university in our society is to obtain increased resources
for its operation and its capital plant, then society must be
convinced that the university serves a useful social purpose.1
To address concerns regarding state support, such as those broached by
former Delaware Governor Peterson, several state legislatures modified the
relationship with their higher education systems in the 1980s. They gave higher
education financial flexibility in exchange for promises of increased productivity
(Ewell, 1990). Now, with constrained resources and the resulting competition for
those resources, it is imperative that public agencies, including public higher
education institutions, document their productivity. This research will investigate
the effects of public support for higher education by proposing and testing
productivity measures.
1 Governor Russell W. Peterson of Delaware, the new Chairman of the Education
Commission, in Sernas, Philip W. "Education Must Prove Worth, Politicians Say," The
Chronicle of Higher Education. Vol. IV, No. 37, August 3, 1970.

In response to the 1980s legislative changes, higher education eagerly
adopted financial flexibility, usually manifested by reduced restrictions on the
expenditure of funds by omitting line item budgets and/or faculty and staff FTE
allocations. However, colleges have yet to reciprocate by formulating or
implementing specific strategies to measure, let alone increase their productivity.
Higher educations hesitation may be due primarily to the difficulty in measuring
public productivity, the result, in turn, of the public sectors diffuse network of
social goals (Arrow, 1993). Assessing the attainment of quantifiable profit goals
provides concrete measures of business productivity, but public sector productivity
is less easy to define or quantify. To that end, this dissertation tests whether an
external outcome measure return on investment in education can document
public higher educations productivity. Productivity is defined to occur if state
support of higher education produces significant increases in income for higher
educations former students. If this research methodology can document
productivity, it can also rank current investments in academic/vocational programs
and services from highest productivity to lowest with important public investment
Public sector productivity has always been scrutinized; American society
has been mistrustful of big government and its bureaucracies since the founding of
the republic (Herson, 1986). This early distrust began with the English
administration of the colonies, continued when America obtained its freedom, and
is reflected most clearly in the Federalist papers. Thomas Jefferson promoted the

idea of volunteer citizen employees to minimize the impact and size of the new
American government. This basic distrust continued as the nation grew. While
American government matured and its functions expanded to encompass more
responsibilities, the public (through its representatives) continued to examine
closely the outcomes of government programs for evidence of achievement and to
demand changes when appropriate. The introduction of the civil service system
and improvements in formulating government budgets were the result of previous
cycles of administrative reform to increase productivity in government (Brownlow,
Merriam, and Gulick, 1937).
The most recent emphasis on productivity reflects increasing national and
state interest in the public sector for two reasons. First, many private sector
businesses have made headlines with attempts to increase their productivity. The
necessity to compete successfully worldwide has forced them to thin their
management ranks and to require the remaining employees to work more. In
concert with the ballooning national deficit and resulting budgetary difficulties, the
examples from business have forced government to apply this distasteful, but
necessary, reduction in size to itself. For example, the U.S. military has suffered
large decreases in personnel because the public questions its size and purpose in
the post-Cold War world and believes that it is a major user of government funds
that could be cut. More recently, Vice President Albert Gores National
Productivity Review moved the federal government in a similar, downsizing
v direction to increase productivity and to cut the budget deficit (Gore, 1993).

Second, the federal budget deficit makes it necessary to verify that citizens
are receiving benefits from federal dollars given to public agencies and private
companies. The recently re-authorized federal Higher Education Act (1993)
contains a section that requires states to establish procedures to review the
productivity of public and private higher education institutions. One review
measure requires that the state assess if the post-education salary shows a
sufficient return on the students tuition investment. Institutions that fail the
review will not be eligible to receive federal financial aid, since they cannot prove
sufficient productivity.
The call for increased productivity is not limited to the federal level.
With the public signaling its disapproval of government spending through revenue
reduction initiatives of which Colorados 1992 Amendment 1 is only the most
recent, state legislators are now even more adamant that state agencies, including
higher education institutions, increase their productivity. Colorado legislators now
require specific outcomes, including productivity improvements, with the clear
intent of linking them to funding increases. Specifically, the 1993 Colorado
legislature passed a bill to provide additional funds for higher education initiatives
in only five areas that the legislature states are important, including productivity
and meeting workforce needs.
Although budget restrictions are significant policy levers, the
reorganization or demise of an agency are also possibilities. For example, the
recent negative vote on the Colorado tourism tax has closed the tourism bureau

because the voting public could find little personal benefit from its efforts
(Mahoney and Lipsher, 1993).
With the threat of budget restrictions and/or restructuring, or even
elimination of public agencies, public sector organizations must develop
productivity goals to demonstrate their viability. Public sector productivity
means effectiveness: the quality and level of service, as well as efficiency, i.e.,
the best use of inputs to produce certain outputs (Dalton and Dalton, 1988).
However, problem definition and implementation are far less straightforward.
The mingling of efficiency and effectiveness results in diffuse and sometimes
contradictory outcomes. For example, an acceptable level of service as
determined by the beneficiaries can result in inefficiencies because of the large
amount of funds necessary to deliver that level of service.
Many groups desire public higher educations services, including recent
high school graduates starting their postsecondary careers, mid-career
professionals pursuing graduate studies, individuals taking a few courses to
upgrade their skills, and those taking classes or complete programs for personal
satisfaction. Delivering all of these differentiated academic/vocational services
productively is a higher education goal, but one that is difficult to quantify and
measure. Each group has different criteria (e.g., access, quality, and cost) for
efficient delivery reflecting its own needs. A 1993 survey of California residents,
supplemented by national survey data, found few concerned about the quality of
higher education but many disturbed about access to and the cost of higher

education (Immerwahr, 1993). Colorado legislators echo these concerns as they
criticize colleges for limiting access by admitting and graduating too few students.
Colorado higher education institutions must document their productivity to their
consumers and legislators, or they too face potential restructuring or downsizing.
This study will assist in developing policy recommendations concerning
state funding of higher education in Colorado. Feasible policy proposals must
consider the empirical data, the political environment, and current practice. This
dissertation, therefore, will examine Colorado higher education productivity
through several lenses. One lens will explain productivity theories and practical
applications as they relate to the public sector and, specifically to higher
education. This lens also includes the value system of higher education, in
particular how that value system influences the calls for increased productivity.
Another useful lens examines current budgeting practices and budgeting
theories. A recent attempt to include productivity measures in budgets was the
widespread implementation of a Planning Programming Budgeting System (PPBS)
at the federal level. Based on specific quantitative goals, PPBS budgets included
several options for achieving these goals (Rosenbloom, 1989). PPBS was not
successful because of the inability to quantify some government goals and many
political barriers to changing the budgeting system (Wildavsky, 1974). Higher
education has rarely incorporated productivity measures into budgets, because of
the inability to control outcomes (Jones, 1989). The difficulties of performance

budgeting in the public sector will be explored, including its effect on higher
A third lens incorporates labor economics to provide a theoretical base.
Human capital theory predicts a persons future income based on current
investment in education (Becker, 1964). Human capital theory hypothesizes that
an independent productivity measure called Return On Investment in Education
(ROIE) calculates the interest received on investing in higher education (Mincer,
1974). A multiple regression equation calculates the ROIE from variables that
influence income.
Governments invest in education for many reasons, such as to strengthen
the economy and to advance technology. When called upon to explain then-
actions, states justify their support of higher education by claiming it propels local
economic growth (Korb, 1991,1992). However, quantifying that effect has proved
elusive (Pencavel, 1991). Using human capital theory, the states return on
investment in postsecondary education is also measurable. ROIE measures the
return on the states investment by comparing the state subsidy for residents with
the increase in those individuals incomes. The third lens will be used to focus on
the research problem statement and results. Those results will then be modified
by the information contained in the first two lens to develop plausible policy
applications. Multiple perspectives are necessary to incorporate the empirical
findings with current political realities and practical barriers. These lenses
provide the theoretical, political, and practical background for this research.

Scope of the Study
Creating a productivity measure for Colorado higher education requires a
quantitative model based primarily on a multiple regression equation. This model
compares the states investment in postsecondary education (including institutional
operating costs and indirect costs generated by governing boards and the
coordinating board) with the return received, measured by individuals increased
incomes. This regression equation will assess the effects of students and schools
characteristics as measured in a statewide student data base on income -- the
dependent variable. The critical independent variable in the model is the states
cost to educate these individuals. This dissertation proposes that the return on
state support is one quantitative documentation of higher educations productivity.
Comparing different returns on degree levels and various disciplines can be used
to develop priorities for investment that could increase Colorado higher
educations productivity.
Students attending Colorado public higher education institutions in 1990/91
constitute the study population. The Colorado Commission on Higher Educations
(CCHE) Student Unit Record Data System (SURDS) data base contains the
necessary student information. The data include type of degrees obtained or type
and number of classes taken, age, sex, race, and ability variables. Each variable
is hypothesized to affect a persons income. The multiple regression equation, in
theory, isolates the effect of each of these elements on income. To test the effect

of the states support of higher education, the equation includes the state subsidy
for each individual in the sample derived from institutional budgets.
Income data are available from unemployment insurance files that include
most wage and salaried workers in Colorado. The wage data are available only
for individuals working in Colorado. This is a critical element, in that the state
has supported many students but only a certain portion will remain in Colorado
and provide a return on the states investment. Former students who left the state
subsequent to their schooling or who do not work will not be included in the data
base. This is not a relevant gap because their income cannot provide a monetary
return to Colorado. Calculating an accurate return requires decreasing the return
by non-contributors. The state also attracts students who stay and contribute to
Colorados quality of life. Those students will also be captured by the model and
will increase the return. However, individuals that were educated in another state
and then come to Colorado will not be included. If one assumes that the higher
education system contributes to the attractiveness of the state, then this benefit is
not included and the resulting return will be understated.
The regression equation will satisfy the goal to approximate the overall
productivity of Colorado public higher education. The equation indicates the
correlation between costs of education and future income, or in financial terms
return on investment on state funds. The return on investment model can also
measure the states return for providing different degree levels and classes. But to
develop a rank for each degree level and discipline area, additional elements are

necessary. To address more fully the total state contribution to higher education,
the model will include the following elements that either have additional financial
implications or address other state goals:
* Capital Construction No one can accurately predict the specific projects
that will be funded in the future for capital construction, since there are too many
variables that affect these decisions (political whimsy, natural disasters, available
funds from other sources, programmatic additions or deletions). Certain
disciplines, however, may be more likely to require higher or lower than average
capital construction funds. The decision to add disciplines with intensive capital
construction needs (e.g., science laboratories) should consider construction
investments as well as operating expenses.
* Financial Aid Although the data on financial aid are by student, it is
not clear how much of the financial aid is returned to the state in the form of
tuition payments. This certainly clouds the return on investment calculation.
Therefore, the costs of financial aid will be included in the cost/benefit model to
recognize more clearly financial aids effect given the ambiguous nature of
financial aid costs as either generating returns on investment or direct revenue
* Access Goals One of the fundamental social purposes of public higher
education is to provide access to under-represented groups. Certain programs may
attract students with low incomes, thereby meeting statewide access goals.

The model will include all of the major subsidies typically provided by a
state to its higher education system to calculate more completely the impact of the
states investment. The model also incorporates a fundamental social goal of
providing access to under-represented groups. After gathering this information
and ranking the investments, this dissertation will formulate policy
recommendations concerning higher education productivity based on the findings
and suggest how the current Colorado higher education budgeting system might
incorporate these recommendations. This thesis may also extend the theoretical
base in the sense of a practical application of the theory; using the theory to
actually measure government productivity has not been the purpose of human
capital theory.
Organization of the Thesis
This study has five major sections: introduction, literature review, research
hypotheses and methodology, findings, and a conclusion that incorporates public
policy implications. After Chapter ls introductory comments, Chapter 2
discusses the substantial research available concerning the theoretical model on
human capital theory and the political, budgetary, and educational implications of
productivity. The next chapter discusses the hypotheses to be tested and their
relationship to the theoretical information presented previously. In addition, this
chapter contains a methodology discussion including an explanation of the multiple

regression equation, data sources, sample population parameters, and statistical
treatment of the data. Chapter 4 presents and interprets results, while Chapter 5
concludes the thesis with the development of the findings theoretic and policy

... even less has happened to answer the larger state and public questions
now emerging about the effectiveness of the higher education system in
ensuring students collectively are capable of contributing productively to
the work force and to society.2
Public sector productivity is coming under criticism as public discontent
with government increases. This criticism stems in part from the international
challenges of the last two decades that forced business to become more productive
and the reform spotlight now shines on the public sector (Kefalas, 1993a).
Society wants public agencies to be more productive, but government employees
need to learn how to achieve that goal within the public service culture. This
chapter reviews three different research areas to answer those demands: (1) the
theories and values underlying public productivity provide the context for making
policy recommendations; (2) public budgeting decision models explain the current
practice for allocating tax dollars to public agencies; and (3) human capital theory
develops the rationale to measure public higher educations productivity. Each
2 Finney Joni and Sandra Ruppert, Assessing College Outcomes: What State Leaders
Need to Know. Education Commission of the States, Denver. November 1991

section discusses the findings in these areas in general and also reviews the
specific applications to Colorado public higher education.
Public Sector Productivity
In a fiscally limited, political climate, funds to provide government goods
and services become more constrained, even as the demand for those services
increases. To continue an equitable delivery of goods and services, agencies must
increase productivity. However, public productivity is a cyclical topic that
alternates between requiring government to be either more efficient or more
equitable (Calista, 1986).
Both efficiency and equity are social goals that government should attain,
but they are difficult to achieve simultaneously. The 1960s saw issues of equity
arise as the government realized that specific groups were receiving unequal
treatment. These inequalities resulted in the groups being unable to achieve their
goals, either in government or in business. Most of the legislation during that
period sought to include individuals previously excluded from basic opportunities;
the passage of the 1964 Civil Acts Right is an example. Political reforms in the
1970s swung to the efficiency side with the introduction of sunset and sunshine
laws, mandatory program evaluations, ombudsmen, and new advisory and civilian
review boards. The inflation rate also contributed since efficiency goals become
more popular during times of rapidly increasing prices (Nagel, 1984). In the

1980s, productivity advocates looked critically at the increasing size of
government in addition to government efficiency.
One origin of this cyclical reform pattern is the business orientation of
American government. As business reacts to changing environments, the private
sector demands that its partner American government change also. Business
principles and rhetoric advocating rational, efficient operations often provide the
basis for public administrative reform cycles (Herson, 1986). Public agencies are
analogous to business firms in that they provide consumers with bundles of policy
goods, mirroring the economic goods provided by the business sector. Just as
business monopolies distort rational, efficient operations in the private sector,
political domination perverts these same objectives in the public sector (Self,
1975). When the public pressures government to be more efficient, politicians
seize the opportunity to berate bureaucrats, but usually do not in reality press for
more efficiency since this could result in some influential constituencies no longer
receiving government services.
The political nature of administrative reform often develops a different
definition than that applied to business. Private and public efforts both attempt to
maximize output with a given level of input. In addition, however, public
productivity also must arrive at an equitable allocation of services and/or a wide
variety of constituencies that receive services as opposed to being purely market
driven. Public productivity, therefore, is both efficient (the quality and level of

service), as well as equitable (the distribution of services) (Dalton and Dalton,
Although the social goals of efficiency and equity are admirable and the
rules provided by business practice can increase efficiency, politics provides both
the impetus and the barriers to improving public productivity. According to Stone
(1988), equity and efficiency are not mutually exclusive. She holds that all
arguments for mutual exclusivity are based on the inherent evils of wealth
redistribution: 1) no one works without the threat of starvation; 2) efficient
redistribution (equity) requires large, inefficient bureaucracies; and 3)
redistributing dollars has unmeasurable, but exceptionally large opportunity costs.
Stone proposes that equal political power for constituencies would simultaneously
achieve equity and efficiency. However, achieving both in the political
environment is challenging because the definitions of efficiency and equity change
rapidly; hitting a moving target is difficult; simultaneously hitting two changing
targets is even more questionable.
The political nature of administrative reform generates different rhetoric
for each perceived problem. The most recent rhetoric about public sector
productivity is generated from the external environment, specifically, international
competition, environmental concerns, and the change from an industrial society to
an information society. These, in turn, pressure public entities towards greater
service delivery efficiencies. Intense business competition generates cries of over-
regulation or bureaucratic red-tape; environmental concerns involve another

interest group, often in conflict with business concerns; and the emerging
emphasis on information requires different managerial skills, specifically in the
area of managing for more productivity, a problem compounded by the emergence
of a more diverse workforce.
Business continues to change from organizations where employees duties
are rote (follow instructions, be punctual, and do the job assigned) to those
valuing employees creativity (understand and enhance the connections between
diverse parts of the company, challenge old information, and generate new ideas).
Once again, the public sector is pressured to mirror its private sector partner.
The public service ethic of equity, statutory authority for rule-making, and civil
service rules to maintain a non-political workforce become or are labeled
obstructions to societys transformation. Barzelays study (1992) of a state
agencys change from a typical bureaucracy to one with a customer orientation
illustrates the difficulties of individuals and processes changing from a culture
based on rules and statutory authority to one of service. Productivity rhetoric
greatly strains the public service ethic in the 1990s, as productivity once again
becomes the byword for both business and government.
Productivity Strategies
Strategies to achieve productivity in both the private and the public sector
are similar: 1) enhance productivity through the use of technology (e.g.,
technology use resulted in a seventy-two percent [72 %] productivity increase in

the private sector), 2) change management approaches to motivate employees
(business estimates 15%-18% increases), and 3) increase the size of agencies or
businesses to take advantage of economies of scale (Mercer, 1992; Bates, 1993;
Pollitt, 1990).
The use of technology is site specific and will change from agency to
agency, but management theory can be applied to many organizations to increase
productivity. The private sector generated most management theories and many
moved into the public sector with varying degrees of success. Those that study
public management advise several strategies to increase public productivity,
including goal congruence, motivational strategies, more specific cost accounting
information, and strategic planning. A few claim that there are no new strategies
but instead all rely on past theories. Pollitt (1990) remarks that many of the
increased productivity strategies regress to scientific management or in his
terminology "Neo-Taylorism." Managers define outputs, measure employees
against those norms, reward those that meet the norms, and maintain tight control
over the entire process.
Others (Bates, 1993; Mercer, 1992) believe that managers must first
engender goal congruence where all agency employees are striving to reach the
same goals. This is similar to decision theory, where after deciding on objectives,
everyone tries to achieve it using optimal means.
Strategic planning also helps agencies clarify and identify their goals and is
an application of systems theory where both external and internal elements affect

an organization and must be included in strategic plans. Productivity rhetoric
contains few elements of human behavioral theory (e.g., employees are unique
individuals and simply not organizational cogs) and the values and ethics of public
service receives little attention.
All of these management strategies recommended to increase productivity
are very mechanistic and de-emphasize the human resources viewpoint. When the
country changed from an agricultural society to an industrial society, Frederick
W. Taylors scientific management theories became the norm. And as the society
changes from an industrial society to an information society, the types of theories
popular in that first transformation are once again favored.
Although Taylors theory does not ignore human needs, managers tended
to implement it mechanistically. Managers held that employees respond to
specific rewards, primarily more pay; to motivate employees to do their jobs
efficiently, managers provide financial incentives. But whether these old theories
can motivate employees who are well-educated, mobile, and less malleable
compared to employees from previous decades is questioned. Denhardt and
Prelgovisk (1992) hold that the purpose of public leadership is to develop new
directions but leaders can only tap those desires and needs already present in their
followers. Therefore, well-educated, mobile employees must agree that improving
productivity is necessary and then leaders can draw upon that understanding.
Decision theory and systems theory are also mechanistic and look at the
world linearly where certain inputs (including the outside environment) produce

certain outputs and feedback mechanisms are necessary. Human resources are
merely one input factor among many, and managers expect them to act logically.
Pollitt (1990) recommends the newest theory culture theory that
interprets the input and output of an agency (or business) through the lens of its
culture. Each entity has a particular milieu that is defined by its environment, its
employees, the goals and objectives of its management, and its industry.
Productivity (or any other valued outcome) must consider the organizations
culture. Culture theory has one advantage over other management theories in that
it does recognize the differences between the public and private sector managers
environments. With that one advantage noted, it is also true that culture theory
has not proven to be predictive, but only descriptive. Many of the companies
described in In Search of Excellence by Peters and Waterman (1982) the
breakthrough publication in culture theory are either in financial trouble (IBM
and Apple Computer) or changed their culture significantly to survive the
changing environment (Hewlett Packard); e.g., layoffs are now common, whereas
before it was an unspoken rule in these organizations that layoffs never occurred.
Several elements of the public service culture must be changed for public
entities to embrace productivity: 1) the philanthropic goal of many public agencies
(Bates, 1993), 2) the conflicting goals of politicians and bureaucrats (Pollitt,
1990), 3) neutral competence (Wilson, 1887) and 4) administration vs.
managerialism (Pollitt, 1990). Philanthropic organizations give to those needing
their resources regardless of cost and exhibit little concern about actual benefit

accrued. Philanthropic government agencies redistribute resources to those in
need and rarely calculate the costs or benefits of that provision. This cultural
view does not recognize the need to document benefits received from its services.
Many politicians have been described as wishing to distribute as much
public largesse as possible to ensure re-election, while simultaneously berating
inefficient bureaucrats. Conversely, bureaucrats want to distribute program
resources equitably, at a certain level of quality, to those most in need of the
goods. This reflects the difference in productivity terminology, where politicians
define it as spreading benefits across constituents, while bureaucrats believe it as
fair and equitable distribution to the most needy. In a resource restrained
environment, both sides cannot attain their goals.
A third element of public service culture neutral competence also is
detrimental to embracing productivity. According to Wilson (1887), public
servants are professionally competent but politically neutral. To stimulate greater
productivity, managers must motivate employees to work towards the same goal.
In the public sector, productivity goals are political, thereby contradicting the
neutral competence dictum and establishing possible role dissonance for public
employees. The most politically beneficial goal may contradict an agencys
statutory authority and its employees professional judgment concerning who
should receive benefits.
Traditionally, bureaucrats administered an agency by relying on statutory
authority and rule-making to maintain objectivity and fairness in dealing with all

constituents. Conversely, managerialism now searches for the best resources to
reach objectives, which was not required of public managers until recently.
These facets of public service culture all weigh against increasing
productivity. One of the common themes running through this analysis is the
dissonance generated by the public service culture cocooned within, yet separate
from, the political environment (Wildavsky, 1992a). This situation is similar for
every new outcome required of the public sector. Heretofore, public managers
achieved changes in the public sector by simply adding new agencies and
employees (e.g., when President Kennedy did not receive acceptable information
in a timely manner from the State Department during a confrontation with the
Russians over Berlin, he simply hired 30 new analysts to provide that and any
future information that he might need [Bates, 1993]). But the current political
environment is unwilling or unable to fund changes to increase productivity. The
political environment now requires agencies to become more productive without
additional funds to purchase new technology or learn how to modify private
managerial techniques to enhance productivity in the public sector or to develop
unique public sector responses to calls for productivity. The political culture of
the public environment requires different actions than the traditional ones used in
the private sector.

Public Higher Educations Environment
The previous discussion of management theories provides a broad overview
of the difficulties in improving government productivity. This section reviews the
specific needs to improve productivity in the public higher education sector,
particularly in Colorado. To develop a framework to discuss Colorados
productivity issues, the following table describes various versions of productivity.
Table 2.1 Alternate Versions of Productivity3
Public/ Private Site of Stimulus Name Description
Both Internal Technical Effectiveness Maximum output related to input
Both Internal Instrumental Efficiency Combination of inputs to produce variety of outputs
Both Internal Organizational Effectiveness Human factors
Both External Allocative Efficiency Approximates market supply and demand
Public External Social Effectiveness Measure of quality and/or level of service to society
Public External Political Allocation Composition of interests served
Table 2.1 illustrates the difficulty in explaining and being accountable for public
sector productivity, specifically the political features of the last two rows. 3
3 Data from Dalton, Thomas and Linda Dalton, "The Politics of Measuring Public
Sector Performance, Productivity, and the Public Organization", Promoting Productivity
in the Public Sector, ed. Rita Mae Kelly, St. Martins Press, New York, 1988, pp. 19-68

Whether deliberate or not, political rhetoric is often ambiguous and differing
regarding the particular version of productivity to be improved. In Colorado, for
instance, legislative definitions vary among allocative efficiency, technical
effectiveness or political allocation, while public higher education continues to
define productivity solely as the quality aspect of social effectiveness (see Table
Adding to this definitional disjunction between legislators and higher
education, the 1992 passage of Amendment 1 (now Section 20, Article 1 of the
state constitution) has significantly reduced the flexibility and the allocation
responsibilities of the Colorado legislature. Combined with expanding demands
for scarce resources from other quarters (such as K-12 schooling and crime), there
are few discretionary dollars in the budget and limited managerial flexibility.
These other needs, declining resources, projections of large increases in high
school graduates, and the requirement to keep workforce skills current, reduce
legislative options to demanding more services for stable or even declining state
subsidies (Policy Perspectives, 1992b).
According to a recent poll in California, augmented by national data, the
public believes that a college degree provides entry to the middle-class and,
therefore, must be available to every qualified individual (Immerwahr, 1993).
Yet, legislators cannot even fund higher education at previous levels let alone
rising ones to ensure the publics growing expectations that everyone who can
benefit attends postsecondary education (Policy Perspectives, 1992a). As a result,

legislators turn to increased productivity in spite of its problems as one of only a
few methods available not requiring adding resources to meet constituents
growing demand for higher education.
Individuals in Colorado public higher education claim improved
productivity (primarily by streamlining administrative systems), but have yet to be
sensitive to productivitys political aspects. The publics clamor for access to
higher education is countered by higher educations assertion that increased access
can only be met with additional funds. This is understandable because the
political aspects of productivity are imposed from the outside and are the last to be
recognized by the relevant organization (Dalton and Dalton, 1988). Although
making progress towards increasing access for all Coloradans, most universities
and colleges remain concerned with documenting and enhancing the quality
aspects of productivity, rather than meeting allocative or political productivity
demands. This activity conflicts with the publics increasing anxiety about access
to higher education (albeit at a reduced cost), while exhibiting little concern about
higher educations quality. (Massey, 1989, Immerwahr, 1993. Hughes, Frances,
and Lombardo, 1991). In summary, the public through its elected representatives
sees the need for higher education to extend access to everyone at the present level
of quality, while higher education will only add new students if current average
levels of quality (which it equates with funding) are maintained.
Higher education institutions struggle to define their role in this new
environment, which, as set out by legislative intent, demands that they:

1) increase productivity by graduating more, sufficiently educated
2) give a higher priority to teaching to increase student learning,
3) meet new enrollment increases without additional funds (Mingle
and Lenth 1989),
4) participate in and develop solutions for community problems
(Rood, 1993).
In addition to being unaccustomed to receiving these kinds of demands, many
institutions lack a clear vision, let alone a plan, that will allow them to meet these
new demands. Their failure in this regard leads to "business as usual," which, in
turn, leads to higher funding requests. These requests trigger, in turn, additional
legislative demands for increased productivity (Policy Perspectives, 1992b).
Some observers actively refute the premise that higher education must
become more productive. Arguments in this vein include (1) the American system
is the best in the world and does not need to change (Policy Perspectives, 1993),
(2) higher educations outputs are too diffuse to be measured accurately (Bowen,
1977), and (3) higher education is too labor-intensive to have similar advantages
in improvements in technology which generated large increases in private sector
productivity (Baumol, 1993). Although these arguments once had a certain
validity, they now fail to persuade.
The American higher education system is considered the best worldwide,
but primarily because of its institutional and student diversity. Most other

countries severely limit the college population to only the top few percent of
academically-gifted students, while American public and private institutions offer
almost any type of training or education imaginable to a wide variety of students.
Higher educations outputs are no more diffuse than many other social agencies
and program evaluation techniques are becoming increasingly sophisticated to
assess many difficult-to-measure outcomes. Granting that higher education is
labor-intensive, improvements in productivity through technology are viable. For
instance, telecommunications can increase access by offering classes to students
who are unable to attend classes on-campus.
Simply arguing against the need to increase productivity is proving to be
ineffective. To maintain current funding levels, public higher education must now
address at least two politically generated productivity requirements. Not only do
legislatures require institutions of higher learning to meet more constituents
needs, they also shifted the responsibility for student success. It is now the
facultys role, instead of the students responsibility, to ensure that students learn
and achieve (Mingle & Lenth, 1989; Policy Perspectives, 1991b). This new
responsibility negates higher educations traditional function of "sorting" students,
or providing access to students but expecting a proportion to be unsuccessful.
Now legislators require that virtually all constituencies achieve their postsecondary
education goals, in addition to just access to higher education, or equal outcomes
have superceded equal opportunity.

Public institutions also must document their programs positive effects on
the community, either by producing a competent workforce or assisting
communities in solving their problems, e.g., K-12 performance, gang violence
(Rood, 1993). These new conditions faculty responsibility for student learning
and service to the community -- require heeding the tenets of political allocation
productivity of meeting constituent needs (See Table 2.1). All legislative
constituencies including individual students, communities, and the general
public must be able to tap the expertise of higher education. Legislators and the
public now see higher education as a strategic investment in both individuals and
their communities. It is no longer the sequestered, privileged bastion of the
highly educated few. Therefore, its traditional role of educating the few is
obsolete. Public higher education must educate all who are capable while being
an asset to its community.
Higher educations hesitancy to meet all of its constituencies needs,
primarily undergraduates needs, is understandable. Levin (1991) contends that
there is no incentive for faculty and institutions to become more productive.
Postsecondary education understands that new funding to match the new students
will not be available and, therefore, any proposed financial incentives will be
small, if offered. To financially support incentives and new programs designed to
meet the needs of new constituencies requires internal reallocation of current funds
by eliminating or downsizing existing programs. The management style of higher
education does not easily accommodate or respond to cutbacks. Higher education

institutions organize themselves collegially and on a decentralized basis, which
prize consensus decision making and results in no specific person be accountable
for an outcome (Policy Perspectives, 1991b) and a management style that does not
have "legitimate" or "coercive" power as in many non-collegial organizations. In
other words, campus administration has very little authority to require faculty to
change (Massey, 1989).
Interest groups comprised of faculty from similar or varied discipline
groups are an integral part of the governance of higher education institutions.
Interest groups generally form themselves around similar occupational or
professional categories (Fischer, 1986); interest groups, such as faculty, with
professional characteristics are usually cosmopolitan in nature. Cosmopolitans
display little loyalty to the organization, have highly specialized (and usually
portable) skills, and seek support outside the organization. Faculty derive their
power from working together to forward their particular aims, which may or may
not complement internal institutional or external legislative goals. When faculty
or other groups do not stand to gain obvious benefits, it is unclear why they
should adopt legislative demands towards increased productivity; indeed, often the
direct opposite is the case.
Productivity as defined by the legislature usually means teaching more
students, but the legislature or the institution must provide an incentive for faculty
to teach more and forego research and/or governance activities. Productivity as
defined by faculty has more elements than teaching, primarily undergraduate,

students. Faculty reward systems at large research institutions usually
acknowledge research prowess more than faculty teaching ability. This reward
system has systematically loosened faculty ties to their institutions (and their
legislatures), and strengthened links to their disciplines (Policy Perspectives, 1990)
leading to faculty moving out of the classroom into the library or laboratory.
Faculty at research universities often want or may be assigned the lowest teaching
load to allow time for research to build their research record. This is an
appropriate economic decision; faculty will be recognized by their disciplinary
peers and recruited by other institutions based more on their research records than
their teaching abilities. Thus, university faculty productivity is increasingly
oriented towards products that educational consumers and legislative funders do
not want to buy (Mingle and Lenth, 1989). This establishes conflicts with
legislatures that believe they have the highest claim on faculty time, while faculty
believe that they must focus more on research to be professionally and
economically rewarded. This is less of a problem at primarily teaching
institutions where a faculty member is expected to remain current in his/her
discipline, but is not necessarily expected to have an active research program.
Other values conflicting with increased productivity are issues of faculty de
facto property rights and the increasing use of support services. De facto property
rights mean that faculty have the use of state resources without owning them.
These types of resources include (1) time to do research, (2) graduate students,
and (3) library and computer resources, and laboratories (Levin, 1991). Use of

the resources allows faculties to meet their own needs and can reduce the time
spent teaching (Massey, 1989). Support services increase discretionary faculty
time by providing many of the services formerly supplied by faculty, like
advising. Collective bargaining terms may specifically state the amount of classes
that faculty teach each year which may further limit legislative desires in this area.
These conditions faculty emphasis on research, de facto usage of
resources, and replacing faculty with support personnel generated a decrease in
productivity as defined by the legislature. Legislators suspicions that higher
education productivity has decreased are not unfounded. Indeed, faculty at all
types of institutions except community colleges teach less than they did 20 years
ago (Russell, 1992).
The previous review of management theories and especially productivity
theory appears to be less applicable to the higher education culture than the public
service culture. Motivating faculty with monetary rewards is not usually
successful, since often facultys value systems hold other non-monetary values in
higher esteem. Facultys requirement for a great deal of personal freedom both
inside and outside of the classroom is exactly what the legislature wants to change,
but precisely what academics treasure most. Therefore, simply redesigning
positions cannot be used as a motivational tool. Moreover, administrators have
little authority to design faculty goals and require faculty to reach those goals. In
fact, many management theories seem to have limited applicability to increase
faculty productivity.

Improving Public Higher Educations Productivity
How can the value system of higher education respond to these new, state-
imposed demands that appear to provide little direct benefit to faculty? Levin
(1991) postulates that to change higher education to value these new types of
productivity will require: a systemic process, with a long time frame, that
provides the freedom to experiment, within the current decentralized management
model. Robert Behns (1992) prescriptions include:
1) Replace bureaucratic rules with a bottom line mentality:
a) choose goals,
b) choose how to measure achievement of the goal, and
c) choose resources to achieve it.
2) Let those closest to the problem fix it.
3) Change the bottom lines to adjust to changes in the environment.
These ideas are simply variations of decision theory where goals are chosen and
results compared to those goals. Behn is essentially telling higher education to
think "businesslike" and choose one or more bottom lines and then achieve those
bottom lines. The previous discussions conclusion of little goal congruence
between the public, legislators, and faculty leads to difficulty in achieving his
ideas. Each group has very different goals based on different value systems and,
to date, little consensus has been achieved on the overriding purpose of higher
education or facultys roles. Without achieving consensus on the goals, then

neither can measurements be chosen to measure attainment of goals, resources
applied, or individuals start implementation. In addition, neither of these
prescriptions appear useful since neither provides incentives or even reasons for
institutions and faculty to change.
In summary, this section discussed the current management theories
relevant to public productivity improvement, various definitions of productivity,
the difference between legislative and higher education definitions, and the values
underlying each. This review helps define the problem and illustrates the current
distance between legislative and educations expectations concerning productivity.
Legislatures, despite their considerable power, actually only have nebulous policy
tools at their disposal to influence the daily output of academic organizations. The
academic organization is immensely complicated because of its diversity. Policy
tools to make one sector improve productivity (e.g. research universities) have no
or harmful effects on other sectors (e.g. community colleges). Constituents
pressures on legislators and academic pressures on faculty are pushing in
divergent, maybe even opposing directions. One way for a legislature to signal
institutions its desires is how funds are allocated and management can signal
faculty in the same manner (Jones and Ewell, 1991). The next section considers
the tangible manifestation of the values discussed previously.

Public Budgeting Models
History of Public Budgeting
Public budgets organize information to support decisions by policymakers.
They are political documents predicated on existing environmental constraints and
the type of decisions to be made determines the budget format (Wildavsky, 1968).
There is no coherent theoretical base, as no one has ever answered satisfactorily
V.O. Keys (1940) classic question "On what basis will be it be decided to
allocate X dollars to Activity A instead of to Activity B?." The decisions vary
based on which of the three major purposes of public budgeting control,
administration, or planning is currently being met. Each purpose uses and
organizes information differently and various budget formats provide the kind of
information necessary for each purpose. The following table shows the types of
information needed for each purpose.
Table 2.2 Types of Budgets
Planning Administration Control
Focus Long-Range Annual Day-to-Day
Detail Very Little Some All detail
Organized by Planning Criteria Outputs Line-item
To support Programs People Purchases
Budget Categories The planning budget has a long-term outlook with
very little detail. This budget assists policymakers to develop policies and make

decisions about alternatives to fulfill the policys intent. It is actually a series of
prioritized values and has the most capability to upset the existing political
balance, since it may rank programs higher or lower than previous budgets
The administrative budget is less detailed, but allows managers to choose
how to implement authorized programs or which person in what organizational
unit has responsibility for the program. This budget provides information for the
annual fiscal cycle and funds work plans for that year.
The control budget is designed to prevent fiscal malfeasance;
administrators cannot spend public dollars above their appropriation or for
unapproved items. These budgets primarily support accounting and purchasing
functions. The control budget is very detailed, monitors the flow of money by
amount and the item purchased, and contains the most detail.
In the early 1900s, budgets consisted of relatively small amounts for which
politicians could hold a specific person responsible. Each public manager would
develop a budget and that amount, as well as the particular items to be bought
would be approved by politicians. Budgets began as control mechanisms and were
quite detailed to prevent the misappropriation of funds, at least in theory. This
type of budget reflected the scientific management environment, where small
details were closely monitored and incorporated strict control of dollars and

The 1921 Budget Act gave the president the authority to develop a budget
for the entire federal government and established the Bureau of the Budget, while
the new General Accounting Office answered to Congress. But the budget format
still controlled the spending of funds and the President received little information
to help him manage the entire federal budget. Not until the 1940s, when efficient
accounting and purchasing systems decreased the need for detailed budgets for
control purposes, did newer budget versions become popular. With "New Deal"
programs, the public began receiving direct, sizable benefits from government and
managers began to document those benefits and the resources required to maintain
the flow of those benefits. The managerial budget became more important as the
demands on and the role of government expanded (Hyde, 1992).
Managers need the type of information that allows them to distribute scarce
resources best. Budgets help managers make four kinds of decisions: purchase
inputs, delegate authority, measure work, and fulfill the purpose of the agency
(Lyden and Lindenberg, 1983). The 1949 Hoover Commission developed the
concept of the performance budget based upon these activities. This type of
budget compared measures of inputs and resulting outputs. Managers could then
vary inputs to increase the quantity or quality of output and the budget provided
the feedback loop for this input/output model. Again, this type of budget reflected
the current management theories where after choosing objectives, a feedback loop
consisting of matching results to those objectives monitored performance.

Planning and Performance Budget System (PPBSt The advent of PPBS
introduced the third type of budget the planning budget. Decision makers
desired information to make choices between policy alternatives and to rank policy
choices. PPBS budgets began with defining certain goals to be achieved and then
developing alternatives designed to meet those goals in various time periods, but
with different levels and types of resources. PPBS budgets included the costs and
benefits of each alternative. According to Schick (in Hyde, 1992), "PPB[S]
reverses the informational and decisional flow. Before the call for estimates is
issued, top policy has to be made, and this policy constrains the estimates
prepared below" (p. 59). PPBS was the first systematic attempt to introduce a
version of performance budgeting into the federal government. PPBS has a long-
term time frame, is output oriented, and tries to develop basic program categories
that may be situated across several agencies. This procedure allows decision
makers to choose between different methods (and agencies) to meet policy
objectives and to allocate scarce resources. PPBS improved upon previous
performance budgets in that it was part of a comprehensive planning system,
where performance was compared to goals, rather than simple input/output
Zero Based Budget. fZBBi A later version of the scarce resources model
is ZBB that President Jimmy Carter introduced in 1977 to the federal government
after using it while Governor of Georgia. It attempted to reallocate funds to
achieve unmet goals or fund new goals without additional dollars. This budget

version is tied to a manager that has decisionmaking authority. The manager first
realigns the agencys projects, ensuring all that remain are necessary and viable,
and then ranks them. Higher level managers rank all lower level managers
projects until all projects for the entire system have been ranked, and request
funds for only those activities at the highest ranks. Contrary to the PPBS system,
this includes political elements. Managers at different levels negotiate to
determine the relative priority of their programs (Levin, 1992). The government
has not widely or consistently used either PPBS or ZBB, because of their political
ramifications. Although both have elements to improve productivity, they also
change the budget process significantly and its associated power bases (Wildavsky,
1986), clearly a political consideration.
Wildavsky (1974) reiterates repeatedly that a budget is a political document
and when budget processes are changed, then the political process is also changed.
The federal government may always rely on the current control budget. Control
budgets preclude looking at the entire budget simultaneously and keeps interest
groups claims to a minimum at one point in time. If few outcomes are stated or
if the budget is "fuzzy," i.e., ambiguous regarding what it must achieve and how,
legislators can develop a rationale for using the funds to fulfill voters needs
unknown or unrecognized when the budget was formulated.
Legislative desires (and possibly agency survival) appear contrary to the
goals of planning budgets that actually choose between goals and policies to meet
those goals, or managerial budgets that choose between programs and

implementation procedures. Any choice necessarily reflects opportunity costs that
lead to something not selected and some constituency and/or interest group being
disappointed. Planning and managerial budgets assist the executive in doing his or
her job, but are not compatible with legislative desires to meet all constituent
needs. They also provide little information about current productivity levels or
how to improve them.
In the era of taxpayer revolts, the cutback budget is popular (N. Caiden,
1990; C. Levin, 1992). The tools of public management, as well as public
finance, assume unlimited economic growth and an increasing tax base. However,
the political climate of the 1980s federal fiscal deficits and tax rebellions ~
resulted in the political realization that public sector revenues may not continue to
grow (or possibly grow at all) at former rates. Current budgeting systems focus
on stability and small, incremental changes and are inadequate for cutting back
(Wildavsky, 1986; N. Caiden, 1990). New systems must be incorporated that
look at a variety of other options:
a) establishing more user fees
b) developing productivity measures
c) contracting out services
d) promoting economic development
Although these seem to be new ideas, the literature has not kept pace with the
actuality. Many government entities have tried these ideas and have had limited
success with them. User fees are considered another form of taxation by those

that pay them, while defining productivity measures remains elusive. Contracting
out of services has had mixed reviews and promoting economic development runs
counter to many environmental concerns. These are simply new names for old
ideas: taxes, cutting costs, and expanding the tax base.
Government Support for Higher Education
According to N. Caiden (1990), "Contemporary fiscal stress apparently
demands an internal revolution in budgeting, revenue raising, and organization of
public services, and the scale and scope of government" (p.240). Caiden
suggests that the theories and applications of economics are the new tools of
budgeting. Individuals rank bundles of public goods as they do private goods and
new budgets need to incorporate democratic participation in developing what will
be funded.
If the public participates in the discussion of what to support, it is
important for the public to understand the rationale for funding public services.
Public services provide two major policy products: public goods and redistribution
of wealth. Higher education is a tool to accomplish both. Higher education
provides entry to the middle class and increases the technological aptitude of the
work force. It benefits not only the individual, but also society by providing the
means for individuals to care for their families and provide the private sector with
well-educated employees.

More specifically, Arrow (1993) contends that governments fund higher
education for two reasons: imperfect capital markets and asymmetric information.
Low income individuals cannot find dollars to attend many public universities, let
alone expensive private schools. Providing education to low income individuals
can result in higher income potentials and a chance to enter the middle class.
Inevitably, public higher education knows more about its products than the
potential student i.e., an information asymmetry. Students usually do not know
which major they will eventually choose or which type of institution or department
best fits their needs; institutions know the type of student that is usually the most
successful at each institution and tries to admit students with those traits. Students
that do not know the types of traits to acquire early in the process are denied
access to a small, private system. But a large, diverse, public system can match a
variety of institutions with a variety of students at a variety of tuition amounts to
increase the probability of student success.
Arrow (1993) also argues that there are three goals of public higher
education: (1) filling the demands of society for trained individuals (production),
(2) the fulfillment of individual talents and of the individual (consumption), and
(3) inculcation of social values (consumption). All of these values have social
worth in addition to individual gain and therefore are eligible to be funded by
public funds.
The federal government provides little funding for undergraduate
education, except for financial aid, but funds research to support graduate

education. The states are the primary funders of the operational needs of public
education. According to Korb (1991, 1992), states fund public education because
postsecondary education initiates economic growth. As discussed previously, one
of governments primary assumptions (to date) is to fund increasing services based
on an ever-expanding economy.
Jones (1989) contends that there are two reasons for state funding of higher
education: to produce educational outcomes, but also to preserve and develop the
capacity of higher education. States are beginning to categorize higher education
as a strategic investment that can solve a multitude of social and technical
problems. Maintaining a higher education system is necessary to a states
preferred or desired quality of life for economic, social, and cultural reasons. In
addition to providing postsecondary education, colleges often provide cultural
events, sponsor speakers on a variety of topics, and faculty may extend their
expertise to the community.
Public Higher Education Budgets Policymakers signal their intentions to
state administrators through budgets (Jones, 1989). The form, amount, and
instructions are the tangible indicators of legislative desires. As stated previously,
there are several different types of budgets and, therefore, different ways to
present information. Allocation at the state level and at institutions usually have
two attributes: to maximize equity and minimize political conflict. Planning and
review policies minimize political conflict by removing discussions of educational
competence from the budget arena (Mingle and Lenth, 1989). Funding formulas

based on input and process relationships (i.e., a defined number of faculty will be
supported based on the level and types of courses taken by students) maximize
equity. Traditionally, the incremental approach (Wildavsky, 1974; G. Miller,
1991) is most common. The coming years budget is generated from last years
budget with variations (almost always increases) due to mandated or unexpected
costs, inflation, or fluctuations in student attendance. The baseline is sacrosanct
(or considered so by institutions) and changes must be additions to that base. This
builds a budget to support particular institutional needs, but rarely meets state
desires (Jones, 1989). It also sends the signal that current institutional
accomplishments are sufficient and no changes are needed, unless of course, the
legislature would like to pay for something new.
Colorados Budgeting Practices
Colorado statute CRS 23-1-105 (3a) states that "Duties and powers of the
commission with respect to appropriations ... budgets for Colorado public higher
education institutions will be formulated taking due consideration of institutions
role and missions, variable and fixed costs, and will emphasize stability of funding
and decentralized financial decisionmaking." This emphasis on stability and
decentralized decisionmaking further decreases the ability of the legislature to
require changes in the current budget to fund new ideas at the expense of old

The current Colorado public higher education budget system is unique
because there is one line item for each governing board.4 Each governing board
distributes dollars to institutions who then assign them to departments and other
units. Many distribution formulas use formulas similar to those used to allocate
state funds to governing boards. The primary driver of all of these budgets is the
amount of student credit hours generated by a governing board, institution, or
academic department. The current formulas fund growth; there are no quality,
output, or performance elements. The greater the number of students taking
classes, the larger the budgets grow for the unit. The current values embodied in
higher education budgets generally thought to be advantageous to institutions are
access (through funded enrollment growth), stability (no budget cuts), and
maximum institutional authority (little direction from the legislature).
New Budgeting Practices
For higher education to be successful in obtaining budget increases, it may
have to change the way it asks for new funds. Rood (1993) contends that number
of students educated or papers published are no longer effective arguments to
request new funds. Arguments meeting legislative requirements are now
4 In Colorado there are six governing boards governing 24 institutions. The
Community College board oversees eleven institutions, the Trustees of the State Colleges
govern four primarily baccalaureate institutions, the State Board of Agriculture manages
three institutions, while the Regents direct four institutions. The University of Northern
Colorado and the Colorado School of Mines each have their own governing board.

necessary or legislators will decide on new objectives to fund. For example, the
Colorado legislature recently passed SB 136-93 that established a new mechanism
to fund higher education through five policy areas. The five policy areas named
for the 1994/95 budget year are enrollment, financial aid, workforce needs, K-12
linkages, and productivity. The legislature defined productivity narrowly as
demonstrating sufficient or increased faculty time spent with undergraduates,
either in teaching or advising. There is a suspicion prevalent among legislators
that faculty at the high profile universities (where their constituents children want
to enroll) do not want to teach undergraduates, but are more interested in their
research grants and teaching graduate students. Productivity measures then, are
one way for the legislature to focus faculty attention and expertise on
undergraduate education.
After legislators define statewide objectives, they must incorporate them
into the budget (Mingle and Lenth, 1991) and refocus the values of postsecondary
education from equity, quality, and access to return on investment (Ewell and
Jones, 1991). If a certain program does not provide revenues to cover the cost of
operating the program, institutions should not offer it, but, instead, propose to
offer one that might generate a sufficient return in its place. The return on
investment model requires a different methodology to develop budget rationales
and faculty hiring. The following individuals suggested ways to develop budgets
based on increasing productivity.

Truxal (1989) holds that grants should be given directly to faculty and
departments, because they are closest to the problem and have the ability to solve
it. However this is perceived to be counterproductive to meeting statewide
initiatives or institutional objectives, since the specific relationship of one faculty
members productivity or a single department to the productivity of an entire
institution, to say nothing of the state, is unclear.
Analogously, Policy Perspectives (1990) suggests that faculty not receive
incentives, but instead departments or divisions receive them, to elevate the level
of policy implementation above the individual faculty member. Again, the actual
productivity of one department becomes problematic to define. What is the
acceptable productivity level for offering service courses or support courses for
specific degrees or course taking for enjoyment without the intent of receiving a
degree? And, of course, the purposes of a given department may not mesh
closely with that of the university as a whole, or, again, that of the state.
Of course, decreasing the size of administration is often mentioned as a
way to increase institutional productivity since monies not spent on administration
would support increased academic budgets. But as explained in the section on
productivity, much of the growth in administration is due to increasing support
services to assume facultys former responsibilities (e.g., advising and tutoring
students). Faculty must reassume those former responsibilities, most likely
without additional compensation or enthusiasm, to successfully increase
productivity by decreasing administrative costs.

As we observed in the previous section, institutions should revise the
faculty reward system to reward for differentiated tasks (Policy Perspectives,
1991b). This essentially means that institutions would need to reward faculty at
an equal or higher level for teaching compared to the rewards for good research
(which in many cases provides its own rewards). Although a teaching reward
system would focus more attention on teaching, it remains contrary to most
disciplinary mores at large research universities. Research publications are
excellent sources of information about the faculty members research prowess and
provide information about areas of faculty research that may fit with another
institutions departmental needs. Although it is possible to evaluate facultys
teaching abilities, there are far fewer fora to publish that kind of information
consistently and easily available to other institutions. Levin (1991) suggests a
cost center approach to calculate specific costs for different services in order to
establish their efficiency, as well as using these data to charge differentiated
tuition rates. High tuition rates would support very expensive programs, such as
medicine or engineering. The drawback of Levins approach is that both
administration and legislators might want the high tuition programs at the expense
of the low tuition program, since more money is always better than less in the
public sector. However, the true measure of productivity is the relationship of
inputs to outputs. High tuition programs that barely cover expenses are not as
productive for the university as opposed to society as those low tuition programs
that generate revenue in excess of costs.

Policy Perspectives (1992b) urges institutions to fund new programs with
monies from discontinued ones, decrease time to graduation, and build effective
partnerships with the community and business to share educational costs. These
suggestions seem simplistic. To fund a new program from a phased-out one, the
one to be phased-out must be viable with students and faculty. If the program has
no students or faculty, then there are no costs to save nor revenue to redistribute.
Tenure delays faculty layoffs and administrative difficulties in choosing between
programs result in few cost savings with this method.
Decreasing time to graduation is contrary to current demands. Businesses
want graduates to know increasing levels of information, as well as workplace
traits (e.g., work in teams, exhibit leadership qualities, and deal successfully with
customers). The private sector must understand that decreasing time to graduation
translates into more training for new employees that is more costly in a highly
competitive environment. Moreover, some students do not intend to graduate in
four years for a multitude of reasons. When students were asked if they intended
to graduate in four years at Colorado State University, the majority answered
negatively. Reasons given included the need to work, wanting to take more
classes than required, and generally liking the college environment (Colorado State
University, 1993).
Building effective partnerships has potential if the partners can agree on
what is to be improved. This is an area fertile for disagreements about what

should be taught. Specific businesses may want certain skills or knowledge taught
by the faculty that the faculty believe do not fit within the curriculum.
Finally, Ewell and Jones (1991) suggest some mechanisms to fund
educational productivity, while leaving the base funding sacrosanct. Incentives,
categorical funding, and competitive grants are all mechanisms that can be used to
change the budgeting system. These are very incremental methods and will have
limited effect when the majority of the funding (base funding) is distributed to
maintain stability and the status quo.
This section reviewed the types of budgets currently available to support
increased productivity. Although none apply directly, some segments are
applicable and with additions provided by the many suggestions for change,
budgets most likely can be formulated to support productivity outcomes. The
political cost of doing so, however, may be underestimated. The next section
provides a proposed basis for measuring higher education productivity. It reviews
human capital theory and how to apply the theory to document higher education
productivity goals.
Human Capital Theory
Economic Productivity and Human Capital
Human capital theory originated with several studies that researched the
components of economic growth. Denisons growth accounting model (1962)

examined income growth while Solows (1957) aggregate production model
studied the elements of economic development. Despite criticisms of these
models, they both generated equivalent answers: increases in human labor and
capital (e.g., land and machinery) accounted for little of the increase in economic
growth. In fact, Solow could account for less than 10% of the economic growth
with increases in capital and labor. Speculations concerning the cause of the
remainder of the growth revolved around the quality of human input, not just
quantity. Solow postulated that technological change was the major driver of the
increase in economic growth.
Using a multiple regression equation, Denison (1962) found that, in the
United States between 1909-1957, real output grew on a per capita basis 2.9%
annually, while inputs grew 1.4% for worker hours and 2.4% for capital stock.
Denison proposed that the residual not captured by the inputs was due to increases
in knowledge (Tallman and Wang, 1992). He estimated that per capita .58% of
the average annual growth was attributable to growth in knowledge, while .67%
per capita was due to growth in education (Denison in Becker, 1964). Using
different methods, each economist found that quantitative increases in capital and
labor alone could not account for the economic growth of the United States.
Research on growth models and quantifying the inputs that generate
economic growth is again popular, but with a new component. Both Denison and
Solow postulated that technological change was an exogenous variable, or a
variable whose value is not influenced by other elements in the model. Current

researchers see technological change as an endogenous variable; other elements in
the equation affect its value. These models specify that human capital formation
generates economic growth because enhanced human capital creates technological
advances. The human capital factor directly affects output through the growth of
technology. Lucas (1988) and Romer (1990) state that in addition to creating
technological change, the average level of human capital accumulation can
positively affect individual human capital and is a positive externality. Or as
Arrow (1993) explains, human capital begets human capital; an environment with
a good base of human capital will increasingly support more human capital.
Romer (1990) postulates human capital affects growth only indirectly; it is a
precondition to developing more efficient capital.
The total stock of human capital, therefore, positively affects economic
growth. The lessening effect of land and capital in the past 40 years in the United
States is due in large part to the increasing influence of a better educated labor
force. Most wealth in the United States today is due to personal wealth or human
capital accumulation and its income effects (Arrow, 1993), whereas land (capital
accumulation) accounted for most individual wealth at the turn of the century.
The Individuals Decision to Invest in Human Capital
While it is clear from the above discussion that human capital, often
expressed in terms of education, improves the productivity of a nation, each
individual must make a decision if she will continue to pursue education beyond

mandatory requirements. One element in this decision has been addressed by
economic theory: the increase in income generally ascribed to education. This
section develops the economic theory and summarizes the empirical results of
human capital investment and its effect on individual earnings.
Economic theory posits two reasons for individuals to "purchase"
education: consumption and investment in human capital. Consumption utilizes
current resources for todays use, but investment uses current resources to
increase future resources. Individuals both consume and invest in education; they
consume education to meet their short-term needs to learn and grow and they
invest to increase future income flows.
Guidelines Schultz (1971) established the guidelines for the theory that
humans invest in themselves to increase their future productivity. This investment
may be in the form of increased health care, more education, or migration for
economic reasons, but he posed that any or all of these investments resulted in
increases in real income.
Human Capital Theory In Schooling. Experience and Earnings (1974),
Mincer showed empirically that education was strongly correlated to higher
earnings. Mincers work demonstrated that wage earnings were a function of
education and experience. Becker (1964) built on this relationship and Schultzs
previous work (1971) by detailing a model of an individuals decision to invest in
education. It is Beckers framework that provides the theoretical foundation for
this work. Becker developed the theory of human capital by postulating that

certain investments in individuals (for instance, education, health care, on-the-job
training, and migration) lead to increased economic productivity. The investment
concept predicts current resources are used to increase future resources, even
though consuming current resources in other ways might result in higher current
Beckers original model predicts individuals will invest in education if the
perceived benefits outweigh the costs; it is assumed that costs paid and wages
foregone because of participation in education in the current period will be
recovered in future periods due to higher productivity rewarded by higher wages.
His initial work was a cost-benefit model that calculated an internal rate of return
on current costs compared to future earnings streams. The original model
Costs=Y' -
1 (1+ry
Becker established the costs of current individual consumption included
both direct costs (i.e., tuition, books, housing, etc.) and indirect costs (wages
foregone) equals the present value (determined by the internal rate of return r) of
the sum of future increased earnings k due to education at j period of time where j
= 1 to n.
Lifetime Earnings The future earning streams are a key theoretical
element. The increase in earnings will be unequally distributed over an

individuals lifetime, but lifetime earnings in total will be affected. In 1976,
Rosen introduced his innovation to human capital theory that initiated the concept
of both the advantages and disadvantages of education and its timing of that
education on lifetime earnings. In equation form:
W(t)= j\{H,C^)emds
W(t) represents lifetime earnings, N is the predetermined length of work life, E
is the current earnings function, s measures time, and r is the discount factor.
Earnings are positively related to accumulated human capital (H), but negatively
related to future human capital accumulation (C). In the model, there are costs
associated with accumulating new human capital and less time to recover those
costs during a shortened lifetime earnings time period. Becker also discussed this
in his earlier work.
Even with Rosens modification and expansion of the lifetime earnings
concept, other economists questioned the validity of the concept that there is a
return to education because of productivity improvements. One question
concerning human capital theory deals with the actual cause of productivity. Does
education increase future productivity or simply screen for other abilities and
conditions? Human capital theory contends that investment in the current period
is rewarded in future periods due to higher individual productivity that is
attributable partially to education. There is another view, however, that holds it is
not the expectation of higher productivity that the market rewards but, instead,

education acts as a screening device. In other words, the market "believes" that
individuals who have more education will have more ability; conversely, those of
lower ability will not attempt additional education. Those with more education are
not more productive because of the education, but because participation in
education signals their higher ability.
Griliches (1963, 1967) studied this effect and found, to the contrary, that
there is a high correlation between wage rates, amount of education, and
productivity. Blaug (1993) also questions the screening theorys ability to support
the outcomes attributed to it. Postsecondary education has a very poor record in
screening individuals. Typically more than half of the individuals beginning in
higher education do not obtain a degree, continuing a trend noticed from the
beginning of the century. Moreover, after individuals graduate, it takes several
cycles before it becomes apparent which individuals are successful. It also implies
the market will not pay for partial completion since there is no signal defined as
obtaining a degree. Current research, although not entirely consistent, has found
returns to those who do not complete 2- or 4-year degree programs. It also
difficult to argue that admission decisions will be precise enough to affect a
persons lifetime income, through many employers and positions.
Although higher education certainly acts as a screen, Griliches research
indicates that education itself also contributes to productivity. With this
background, investment in education can then be examined as a factor in future
productivity and the return on investment paradigm is legitimate to this purpose.

Shape of the Lifetime Earnings Profile The shape of the lifetime earnings
curve is determined by the experience variable. Mincer (1974) researched the
effect of experience or age on human capital. He found experience to be a crucial
variable that had a significant effect on income equal to educations effect. He
also found that experience, instead of age, was more highly correlated to income
because age may or may not indicate work experience. Figure 2.1 shows the
lifetime profiles of individuals with a high school education and one with a college
The important information
illustrated in this figure is the
relationship between the lower wages
of the college graduate while attending
school compared to the higher wages
achieved after graduating. The return
on investment theory labels the
difference in income between the high
school and college graduate as the
return resulting from the investment in college when income would be lower than
the high school graduate.
There are two reasons that explain why experience generates increased
wages. One is human capital theory that postulates as one gains experience, one
becomes more productive. The other the implicit contract argument states
Figure 2.1 Lifetime income profiles by
education 1990 Colorado census data
graphed by author

that employers pay new employees less than their productivity would indicate and
older employees more than their productivity requires. This is the implicit
contract between employee and employer and is designed to discourage expensive
employee turnover. However, studies (e.g., Griliches, 1964) have found that
productivity does increase with experience.
One possible reason for the conflicting results pointed out by Becker
(1964) who distinguishes general training from specific training. When an
individual receives general training (e.g., a liberal arts baccalaureate education),
this training can be applied to any firm in any industry, while specific training
(e.g., sales training for a firms product) applies primarily to the firm providing
the training. Obviously, firms prefer to provide specific training to increase
productivity and decrease turnover, while general training is more valuable to
employees since it is applicable to more than their present position. The type of
training received may have influenced the differing results of these studies. Jung
and Magrabis (1991) research results indicate that employers offered individuals
who changed jobs but could transfer skills developed on previous jobs higher
salaries than those offered to individuals changing types of positions or industries
with similar levels of job experience. Shaw (1984) has developed occupationally
specific experience variables to test the effect of very specific experience variables
on income to further define experience variables.
Developing an experience variable is difficult because few data sets contain
the information. Typically, researchers developed and used a proxy variable for

experience: for instance, Age minus Schooling Completed minus 6 Years. This
is acceptable for individuals that start working full time after school and continue
until they retire. This working pattern is more typical of white males than other
gender and ethnic groups and using this proxy may partially explain wage
Shifts in the Lifetime Profile Experience is the primary driver of the
shape of the lifetime earnings profile, but other variables can shift the curve up or
down. Ability, race, gender, and different educational levels are discussed in this
context. It is important to measure the difference between lifetime earnings with
and without the relevant education. Education will only enhance a persons
earning capacity by adding to his/her ability to earn, but it does not predict total
earning capacity.
Income Colorado 1990
HS Grads 1 College Grads Whites
-H- Blacks -"S- Hispanics Other
Figure 2.2 Income profiles by education and race -1990
Colorado census data; graphed by author.
Figure 2.2
shows the mean
money earnings of
males in 1987. At 22,
when the college
graduate enters the
workforce, his/her
earning curve rises
much more steeply
than that of the high

school graduate as a result of the productivity gains generated by higher
education. Thus, lifetime earnings will be consistently higher than the high school
graduates earnings. Ethnic affiliation also shifts the lifetime income profile as
illustrated in Figure 2.2.
Ability After Becker developed one of the first methodologies to
calculate an actual return on investment, several criticisms of his model emerged.
Critics claimed that education was not the only element that contributes to wages;
they questioned if innate ability were not equally, or even more important. And
leaving ability out of the calculation seriously skews the results. Ability and
education are highly correlated and omitting the ability variable, in addition to not
specifying the relationship correctly, usually acts to increase the value of the
return to education. Griliches and Mason (1972) found that the returns to
education were overstated by 12% by omitting the ability variable.
Much of the research in the 1970s-1990s refined the ability variable, which
was seen as having two components: innate intelligence and environmental factors.
Researchers tried many variables to predict the effect of ability on future wages,
including IQ tests, family income, and fathers occupation. Using brothers
(Olneck, 1976; Jencks and Brown, 1977; Chamberlain and Griliches, 1975) to
lessen the disparities in innate genetic ability eventually led to researching
monozygotic twins (Taubman, 1975; Behrman and Taubman, 1976; Ashenfelter
and Krueger, 1992). Theoretically, identical twins have identical intellectual
ability and very similar environmental experiences that control for ability, thus

allowing the research to focus exclusively on education as a predictive variable.
Although these studies greatly expanded knowledge concerning the impact of the
ability variable, the methods used to gather the data self-reported education
status and income levels led to validity questions.
Recent studies use instrumental variables either to omit errors inherent in
ability variables or to estimate their biases. Ashenfelter and Krueger (1992) used
one twins reported schooling in the second twins multiple regression equation
since twins should have very similar ability, but ones schooling should have no
effect on the others income. Predictably, they found that return to schooling was
underestimated because of the difficulty in measuring ability. Blackburn and
Neumark (1993) believed that IQ and test scores have substantial errors when used
as ability variables in retum-to-education formulas. This study used IQ and
Knowledge of the World of Work (standardized ability test) as instruments for
each other to run the ROIE equation. This study specifically reviewed industry
and occupation income differentials to test if the differentials could be attributed to
ability, instead of to education. They found no evidence to support that
hypothesis; education does affect ones wages regardless of the industry and/or
The very high returns to education documented in the 1980s have been
attributed to industry shifts and returns to very high ability individuals (Blackburn
and Neumark, 1993; Acs and Danziger, 1993). Both research studies found high
ability individuals earned very high returns on their education.

According to these studies, the ability variable is necessary for the
specification of the returns to education equation. The difficulty is to develop an
ability variable that is easily obtained, can detect the difference between genetic
and environmental influences, and is statistically valid.
Gender Beckers emphasis on urban, white, male college graduates
reflected the majority college population at that time. Many women did not enter
the workforce or did so sporadically, and minorities had many educational and
economic opportunities denied them solely because of race. As both of these
populations entered the workforce as college graduates, research in the 1970s and
1980s often examined the impact of gender and racial discrimination (Nord,
According to human capital theory (which Becker developed in 1964),
females earn less than males because they have a weaker attachment to the work
force. Females leave the workforce to care for families, enter and leave the
workforce periodically because of type of jobs or family moves, and work more
frequently in part-time positions (primarily teaching) that depress wages. Smart
(1991) found that female academics had lower salaries primarily because they had
lower academic ranks as a consequence of having less experience than their male
counterparts. However, this differential appears to be declining; Wellington
(1993) conducted a longitudinal study of male and female workers from 1976 and
1985 and found a 4% decrease in the wage gap, primarily because of the average
increase in female work experience.

Experience is a legitimate reason for some pay differentials between males
and females, but there is also evidence of institutionalized sexism. In one study
(Barron, Black and Lowenstein, 1993), females were offered positions with lower
pay, less training, and that required using less capital than males. Employers
were less willing to invest in female employees than in male employees. The
authors postulated that employers believe women exhibit a weaker attachment to
the work force and therefore will spend less resources training females or
allowing them to use easily damaged capital. Of course, an opposite explanation
may be true. If women are not offered the same opportunities, they may change
jobs more frequently looking for better working conditions.
Becker also argued the theory of compensating differentials applied to
women: women, with their family responsibilities, would be willing to trade
lower wages and fewer benefits for more job flexibility. Seccombe and Beeghley
(1992) found that females did not have more flexibility in their jobs; in fact they
often have less flexibility because of the types of positions clerical, sales
many women occupy. Another reason given for the wage differential is that, on
average, women possess less mathematical ability and the market rewards scarce
resources, in this case, math competency, with higher pay (Paglin and Rufolo,
1990). Women, because they have no comparative advantage in occupations that
require quantitative abilities, choose occupations that require little or no math and
therefore are accordingly compensated less. This research ignores cultural
conditioning concerning mathematical ability and interest.

The wage gap is much smaller between entry level, usually young, male
and female workers; in 1991, the mean salary for 1990 female graduates was 87%
of the mean salary of 1990 male graduates. That gap may be due in large part to
the fewer females in high wage occupations like engineering. The gender wage
gap may continue as an artifact of the time period when females earned
considerably less than males, until older, underpaid workers leave the workforce.
It appears that the lack of similar years of experience is the most telling
explanation of the gender wage gap. The gender wage gap may be an exogenous
variable (females choosing for a variety of reasons to enter and leave the
workforce), or an endogenous reason (structurally, employers offer females less
pay because they are females). Since females make less on average, and thus
attending college is less costly for them because of lower opportunity costs, the
theory cannot predict if the private return to women is less or more than the
return to men.
Ethnicity Becker also examined wage records for nonwhites and found
that return on education was much less compared to the white return, because the
wage differential between those with a college degree and those with a high school
education was much less. This was due, according to Becker, because their costs
were much lower; nonwhite high school graduates earned much less and,
therefore, attending college accrued lower foregone earnings. In addition, the
expense of going to college was less because nonwhites usually went to lower
quality schools than their white counterparts.

During the 1980s, however, the disparities increased for different reasons.
Acs and Danziger (1993) found that mean earnings of white male college
graduates increased (7% to baccalaureate recipients and 10.5% to holders of
masters and doctoral degrees) between 1979 and 1989, while mean earnings for
baccalaureate educated black males increased by much less (2%) and Hispanic
males experienced no increases. In those 10 years, mean earnings for blacks with
graduate degrees decreased 3 % and increased 1 % for Hispanic males. This
finding may be compounded by individual choices, since many minorities pursue
graduate degrees in disciplines with lower pay, e.g., education.
Several researchers studied the effect of discrimination on wages and
higher educations ability to decrease that wage differential. Nords research
(1986) found that black males and females received a larger return to a college
education for workers under 56 years of age than whites. Using a decomposition
wage model, Nord found that approximately 75 % of the differential between male
wage rates is due to discrimination against blacks. The inability to work in
certain industries or at certain positions contributed to black male discrimination,
as well as living in the South. White females received lower pay compared to
black females. Wellington (1993) found that on average, black females have more
experience than white females, which may affect this finding. Both black males
and females decreased the wage disparity by attending college. ONeill (1992)
found that the wage disparities between black and white males had significantly
narrowed from 1940 to 1980 but had widened again in the 1980s. She postulates

that this disparity increase was due to blacks having lower skills (as indicated by
standardized tests) than whites with identical education levels. Until the skill
levels are comparable, there will continue to be wage disparities.
Empirical Specification and Results
Multiple Regression Equation All of the above variables affect an
individuals wages. To isolate the effect of education, all must be included to test
educations effect (Mincer, 1974). Expressing these variables in a regression
InY = a + jSE + yA + AH + 8S + OR + e
in which In Y denotes the natural log of income. Income is dependent on an
individuals education (E), intellectual ability and environment (A), experience
(H), his/her sex (S), race (R), and undefinable random differences or residuals (e).
The natural log of income is used because there is a positive skew to income of
individuals with postsecondary attendance, especially if those with graduate
degrees are included. Taking the log of income produces a more normal
distribution. Using natural log of income results in an equation that calculates rate
of return (instead of absolute dollars) and in which the beta (0) coefficient
estimates the percentage increase in income caused by an increase in education.
Researchers traditionally specified the education (E) variable as the number of
years of school attendance. If the specification of the ability, education, and
experience variables have captured their effect on income, then the residual is

expected to be normally distributed with a mean equal to zero and estimates are
Empirical Estimates of Returns to Postsecondarv Education There are
numerous studies on the returns to college education, each using all or
combinations of the variables described above. The following table shows the
various returns.

Table 2.3 Returns to Education Measured by Years of Education
Study A 19735 B 19736 B 19747 C 19928 9 9E 1993 F 199310
High School 21% 16% 37.9%
$283 7% 16% 8.3% 5% (Classes) 9.9% (degree)
College BA/BS $340 13% 12.7 16% 29.1% 5.7%
Graduate Degree $194 (MA) $633 (PhD) 5% NA 16% 13.5% 11.9%
R2 .15 NA NA NA .282 .2010
5 Wales, Terence. "The Effect of College Quality on Earnings: Results from the
NBER-Thomdike Data" Journal of Human Resources. Summer, 1973, pp.306-318.
Numbers in this column reflect increases in monthly earnings.
6Eckaus, Richard. Estimating the Return to Education: A Disaggregated Approach.
Carnegie Commission on Higher Education, New York, 1973.
7 Becker, Gary. Human Capital. NBER, New York, 1974
8 Ashenfelter, Orley & Alan Krueger. Estimates of the Economic Return to Schooling
from a Sample of Twins. No. 4143. NBER, NY. 1992. Sample equals 300 twin pairs.
9 Acs, Gregory and Sheldon Danziger. "Educational Attainment, Industrial Structure,
and Male Earnings Through the 1980s", Journal of Human Resources. Summer, 1993,
pp. 618-641.
10 Kane, Thomas & Cecelia Rouse, "Labor Market Returns to Community College",
Unpublished, October, 1992.

The previous table illustrates the effects of different specifications and
methodologies, but, in general, we can deduce that the rates of returns are larger
for high school and the rates decrease for each higher level of postsecondary
education. One of the unclear areas is in the "Some College" row. It produces
very low returns compared to the other entries, but many older studies did not
discriminate between receiving a certificate or associate degree and dropping out
of a baccalaureate program. It is unclear what many of these studies were
measuring on this line.
The research of Becker,
Wales, and Eckaus was done in the
1970s with information from
previous decades. The 1990s
studies show much larger returns.
This is probably due to the large gap
between earnings of high school
graduates and college graduates that
occurred during the 1980s. Murphy
and Welch (1989) found a 70% gap for average salaries between these two groups
in 1986 and found that the return to individuals within five years of graduation
into the workforce actually exceeded the average return for college graduates
(Figure 2.3). This is a significant departure from the past 25 years, in which first
entrants compared to experienced workers

entrants saw little return until they accrued experience. This large return to
current labor force entrants could skew current human capital studies since it is
difficult to predict how lifetime earnings will relate to entry salaries. Previous
research will provide few guidelines to predict lifetime incomes for new workers.
Degrees Many studies calculated the return to different levels of college
schooling. There is almost universal agreement that baccalaureate and graduate
degrees will generate a return with the highest returns accruing to baccalaureate
degrees. This result is compatible with the theory of diminishing returns.
The results for sub-baccalaureate degrees, however, are mixed. Recent
research provides little support for returns to certificates (usually 1 year degrees)
and mixed results for associate degrees (2-year degrees) (Grubb, 1993; Stevens,
1993; Kane and Rouse, 1992). An associate of arts or associate of science degree
comprises the first two years of a baccalaureate degree, while a vocational
associate degree prepares someone to enter a trade. Kane and Rouse (1992) found
a 11% negative return to associate degrees for men, but a positive return for
women (primarily because nursing degrees provide a high return and most nurses
are female). Grubb (1993) found no return for either certificates or vocational
Different types of degrees also generate different returns (Korb, 1991).
Most studies have found that professional degrees in business, engineering, and
medicine generate the largest return, while the humanities generate the smallest
return (Taubman and Wales, 1974; Eckaus, 1973; Grubb, 1992a; Korb, 1992).

Classes Human capital theory predicts that returns to various
academic/vocational offerings differ based on uneven innate abilities and different
types of disciplinary programs. The research on return to education without
obtaining a degree is mixed. Grubb (1992) and Hollenbeck (1991) have found
negative returns to class taking without obtaining the degree, while Lyke, Gabe,
and Aleman (1991) found positive returns. The research for these studies applied
to two-year degrees and classes only.
The results of three studies (Grubb, 1993; Acs and Danziger, 1993: Kane
and Rouse, 1992) published recently concern the return to credits earned at both
two-year and four-year institutions. Grubb (1993) found a return to four-year
business and engineering credits only. Individuals of all races received negative
returns for completing less than a baccalaureate degree, although it was unclear
whether these individuals dropped out of baccalaureate programs or received
certificates and two-year degrees (Acs and Danziger, 1993). Conversely, Kane
and Rouse (1992) found approximately a 5 % return for every 24 credit hour block
completed at either 2-year or 4-year schools.
Some studies show an increasing return from postsecondary education for
each additional year attended. Ashenfelter and Krueger (1992) found a 16%
increase in income for every year completed, but the finding seriously deviates
from previous studies that found the largest returns to elementary and secondary
education with increasingly smaller returns to higher levels of postsecondary
education. If one accepts Ashenfelter and Kruegers findings, those that complete

less years earn lower incomes, which is an indirect indication that taking classes
(without obtaining the degree) is worth less in the labor market.
Linking the Individuals Decision to Social Rates of Return
The individual makes the decision to enroll in postsecondary education
while the governments role is to provide incentives or disincentives to do so.
Thus we need to ask: What is the justification for government to spend tax dollars
to provide incentives that make postsecondary education attractive to individuals?
Costs to Society Beckers model estimates the private return to education
of the individuals investment. He also calculated a social return with slightly
different data sources, but using the same model. The social returns to education
are more difficult to calculate since all social returns are not captured in
individuals incomes. For example, the Lucas (1988) model postulates positive
externalities due to human capital accumulation accrue to the individual but also
accrue to society since the level of total human capital has increased which is
associated with positive economic benefits.
Becker held that there are both direct and indirect social costs; the direct
social costs are the public subsidies paid to public institutions and should include
capital as well as all educational operating costs. To develop the indirect social
cost figure, Becker defined indirect social costs as books, living expenses, and
foregone earnings. Individuals pay the majority of these costs (although some are
covered by financial aid subsidies) and, therefore, they affect the calculation very

little. Using differences in income as the dependent variable, public subsidies as
the direct social costs, and earnings foregone to estimate the indirect social costs,
Becker calculated that the social return equals 13% for the 1939 cohort of urban,
white, college educated males and 12.5% for the 1949 cohort of urban, white,
college educated males. The private rate of return for these cohorts was 14.5 %
and 12.7 % respectively. Other researchers (Taubman and Wales, 1974) have
further documented that the social rates of return are slightly less than private
Although Beckers specification of the model to include foregone earnings
may have been valid for the 1939 and 1949 cohorts, it is no longer convincing,
largely because of demographic changes in the student population and their higher
education patterns. Nationally and in Colorado, a large proportion (approximately
75 %) of students are adults, most of whom continue working. At Colorado
community colleges, more than half of the individuals taking classes take only
one. This phenomenon questions the validity of adding foregone earnings for
many, if not most, students to a social return equation. There are still 18-year old
students who are attending school full-time and are not working (or only working
part-time). But these represent a much lower percentage of students (25 %), and it
is less clear if foregone earnings should always be included given the current
student population.
Social Benefits Governments invest in education because the level of
labor force education is perceived as a deciding factor in the labor forces ability

to react to new technology, develop new methods and processes for manufacturing
and service industries, and take advantage of new opportunities (Weisbrod, 1962;
Tizcinski and Randolph, 1991; Creedy and Francois, 1990). While these are
germane to all levels of education, these qualities are most often attributed to
postsecondary education.11 Governments invest in postsecondary education
primarily to foster those qualities, which lead to a more productive economy.
Other Social Benefits Given that education does appear to contribute to
future earnings, increased salaries and wages are only one of many benefits from
education. Other, noncognitive, largely-social benefits increased awareness of
societal issues and political participation, consumer sophistication, positive
intergenerational-effects are also thought to be enhanced by education.
Although these benefits are augmented by education, they are largely
unmeasurable and any calculated social benefit of education is admittedly
understated because these benefits are absent from the equation.
A recent study by Grubb (1989) measured some of these noncognitive
benefits associated with postsecondary education unemployment status,
occupational status, job satisfaction, overall satisfaction, voting record, political
participation, social participation in community organizations, clubs, service
organizations. He found that all supposed benefits, except political participation
11 State investment in K-12 education is often explained by the desire to have a
literate population. The high returns to K-12 education are due to the acquisition of basic
literacy and numeric skills.

and voting record, were not statistically significant. Individuals with vocational
degrees, however, did not demonstrate even increased political participation or an
enhanced voting record. Therefore, the return to all degrees with the exception of
vocational degrees, may be slightly understated by omitting the positive impact of
increased participation in politics and voting.
Another study in the African republic of Malawi supports Grubbs
conclusions. Graduates of four different types of institutions (Nursing,
Agriculture, Polytechnic, and Arts and Sciences) were surveyed to measure the
contribution of their education to various elements, including educations
contribution to the graduates participation in activities to serve their communities.
The average for this measure was the same for all programs (Dubbey, Chipofya,
Kandawire, Kasomekera, Kathamalo, Machili, 1991). Of course, the applicability
of the Malawi results to the Colorado population is unknown.
Nor are noncognitive retums/benefits subject to differentiation. Lau (1979)
argues that as the level of education increases, the relative importance of
noncognitive benefits declines, as opinions and values become more set. Ladd
and Lipset (1975) found significant differences in political views for students in
different academic fields, but those entering the academic field held the same
views as those already studying the discipline. Different academic fields may
attract individuals with different political stances but the disciplines themselves
may have no impact (Bowen, 1977). Therefore, noncognitive benefits do not
appear to be a statistically significant result of a college education with the

possible exception of political participation and voting. Indeed, the presence of
one of the highest education rates among democratic nations juxtaposed with one
of the lowest voter turn-out rates (in federal elections) might give one pause
regarding the correlation of these variables.
Migration Part of Beckers (1974), Shultz (1961) and Mincers (1974)
definition of human capital investment was the ability of individuals to migrate to
find better economic opportunities. Trzcinski and Randolph (1991) found that
younger individuals (less than 40 years old) were the most likely to migrate.
Therefore, it is likely that if certain types of degrees or classes attract younger
individuals, individuals with these degrees (or classes) will exhibit higher
migration rates, which is a source of potential bias. According to a 1959 study
(Sharp, 1970), male baccalaureate graduates in fine arts, English and religion
majors were the most mobile, followed by natural science, engineering, and social
science majors. Least mobile were business, education, and health majors.
Migration is a positive individual investment strategy, but it is clearly not
in the best interests of the state that partially paid for the education. The state-
educated individual moving out of state will not contribute to its economy or
social environment in any but the most indirect way (e.g., developing a new
technology that benefits the national economy). Indeed, by removing a portion of
the states investment portfolio (the individual student), the state is surrendering an
asset. It is, therefore, of concern to the state if there are certain disciplines that
routinely produce little or no value to the states economy.

Conversely, migration into the state is positive. Attracting college-
educated individuals has been a long-standing policy of Colorado. Colorado has a
very highly-educated population compared to other states and much of its
population was educated outside of the state. However, with the large growth in
Colorados population, the benefits of migration may be less positive when the
disadvantages (e.g., overcrowding, crime, pollution) are also considered (E.
Miller, March 19, 1994). If migration from the state is an area of concern, then
it is necessary to balance that concern with a review of higher educations ability
to attract individuals permanently to the state.
Similar Research Studies This research will use the theoretical basis of
human capital theory to test whether the states investment in postsecondary
education has a measurable return, as well as rank returns from different
academic/vocational programs. In this vein, Tolley and Olson (1971) researched
the relationship between individual average income and educational expenditures
per pupil in the K-12 system. Because expenditures in K-12 are dependent on
local school district mill levies, many of the independent variables in the equation
reflect that qualification: private property owned per employed person, population
per square mile, urban percentage, percentage of nonwhites, and number of pupils
per employed person.
In spite of this different specification, their findings still remain relevant to
this research question. Their research found that expenditures per pupil were
significant at the .95 level. They also specified the equation to test if average

income affects per pupil expenditure and found that income per employee was a
significant indicator of expenditure per pupil. However, income affected
educational expenditures (a coefficient of .78, significant at the .99 level) much
more than educational expenditures affected income (a coefficient of .259,
significant at the .95 level). This study used 1960 census data for the 48 states to
develop its variables. It found a positive, statistically significant relationship.
A major flaw of the Tolley and Olson study is that it is unreasonable to
expect that all employed adults in a state were educated by public and/or local
elementary or secondary school systems. It makes more sense to expect that
average income affects educational expenditures, since income determines
ownership and cost of property. This study exhibits a preliminary foray into
researching the effect of government support of education on future income.
Related Concepts from All Three Approaches
A review of the relevant theory bases indicates that these three concepts
support the proposed research:
a) The public sector, specifically higher education, is strongly urged by the
public and their elected representatives to assimilate productivity as a goal,
and to foster that goal, the state must review its resource allocation system
to reward systematically a credible measure of higher education

b) Current managerial theories and budgeting practices appear inadequate
to support productivity improvements within the contemporary higher
education culture.
c) Economics is one avenue to provide a credible measure and allocate
resources to ensure productivity (Ewell and Jones, 1991; N. Caiden,
d) Human capital theory was formulated to explain productivity
improvements (Mincer, 1974; Becker, 1964). Current needs to examine
government productivity as it relates to higher education seem a natural
extension of this theory.
Therefore, the three areas discussed are relevant to developing interdisciplinary
policies to address government productivity questions posed by the legislature and
the public. The results from the ROIE review are the basis for defining higher
education productivity: if students receive higher incomes because the state
invests in public higher education, then higher education is one way to meet state
goals and is productive i.e., "well-spent". The following chapter develops research
hypotheses to answer these questions and explains the methodology to test the

The research discussed in the previous chapter found a significant
relationship between a students investment in education and future income.
Human capital theory suggests a relationship between the states investment and
future income. The following figure illustrates higher educations productivity by
measuring its former students contribution to the state economy; this model
compares state investment to the income flows received by former students.
Figure 3.1 Revenue streams supporting public higher education

There are three main revenue streams in postsecondary education: an
individuals tuition and fees; the state taxpayer who supports college programs
with tax dollars; and the college that raises money externally from alumni,
businesses, and generates research dollars (the area enclosed by the dotted line in
figure 3.1). This research focuses on the state taxpayers contribution to
postsecondary education. There are several types of costs funded from tax dollars
financial aid, operating costs, and capital and all will be included to measure
the return on the states investment in higher education. The return on the states
investment then provides a quantitative measure of Colorado public higher
educations productivity in terms of the return on the states investment.
Pushing this idea to the limits, the most productive use of state funds might
be to rely entirely on the private higher education sector to provide all of the
states higher education needs or other states. This is a false argument for
Colorado. There are only three non-profit, private, four-year schools and the for-
profit, baccalaureate and graduate schools (e.g., Nova and University of Phoenix)
offer only a few disciplines, primarily business and education. The proprietary
schools (for-profit, vocational schools) are numerous but very expensive. To
extend fully access to Colorado residents, as desired by the legislature, requires
large increases in financial aid and financial incentives to increase the number of
programs available and would relinquish the minimal control of higher education
that the legislature now possesses.

Assuming that public higher education can become more productive, the
proposed model measures higher education productivity with the following
method. First, a regression equation generates returns based on the same
variables discussed in the literature review:
InY = ce + jSE + *yA + AH + 5S + 0R + e
These returns then are modified by other factors: (1) the percentage of low income
students higher education attracts, (2) average capital construction costs and
financial aid by discipline, and (3) the percentage of former students leaving
Colorado, while those students attracted to and remaining in Colorado are
included in the original regression equation.
Primary Hypothesis
The productivity of the states investment in public higher education is
unknown at this time. But the legislature recently declared that higher education
must meet two goals relevant to this research: (1) provide the state economy with
a work force trained for desired industries, and (2) increase faculty and
administrative productivity (SB 136 policy areas). The primary research
hypothesis measures the increase in income due to education to approximate public
higher educations productivity and contribution to the states economic climate
and indirectly indicates that higher education is meeting the states goal for a
trained work force.

The states investment in public higher education will
exhibit a positive return to the state.
Secondary Hypotheses
Individual disciplines and/or degree levels should exhibit a variety of
returns based on the varying levels of state investment in and varying incomes
accruing to different disciplines. To rank disciplines from most productive
(measured by high returns) to least productive, this research project will test a
secondary set of research hypotheses.
H2a: There will be significant differences in the
returns on different disciplines.
H2b: There will be significant differences in the
returns on certificates, a two-year degree, a four-year
degree, and graduate degrees.
H2c: There will be significant differences in the
returns for individuals that only complete classes
compared to those that finish degrees.

To test the primary research hypothesis, a standard multiple regression
equation predicts the strength of the relationship between variables, in this case,
between aggregate state investment and individual student income. Regression
analysis shows relationships, not causality (Tabachnick and Fidell, 1989), while
the theory hypothesizes causality (direction of the relationship). Human capital
theory predicts that current investment in education will increase future income
and the regression equation will measure the presence and intensity of that
relationship. The regression analysis will include two types of state investment
operating costs and governing costs to measure more completely the states
expenditures to educate individuals. It will also include other variables that
previous research (Mincer, 1974; Nord, 1986; Ashenfelter & Krueger, 1992) has
shown to affect income, including sex, race, experience, and ability.
To test secondary hypotheses 2a and 2b, each record includes identifiers of
degree level (certificate, associate, etc.) and area of study (e.g., education,
engineering, etc.). After controlling for these identifiers, the regression equation
calculates b coefficients for each discipline and each level to compare the
productivity of each discipline and each level. Korb (1992) has found significant
differences in returns for individuals holding different types of degrees and
different disciplines. However, that research defined the education variable as
years of education accrued, instead of state costs. This method provides a much

wider range of costs because of the large variations in the state subsidy to
different disciplines and degree levels.
The final secondary hypothesis will test differences between the b
coefficients on the cost variable for those individuals that obtain a degree in a
discipline and the cost variable for those that only take classes in the same
discipline at the same level. There are mixed results reported in previous research
(Grubb, 1993; Kane and Rouse, 1992) and no research found to date actually
compared the return on the two methods of receiving instruction.
Return to Education Regression Equation
The primary and secondary hypotheses can be tested with the same
multiple regression formula:
InY = a + /3EEd + ADE^ + 7A + AH + /3H2 + 5S + 0R + e
in which
In Y = natural log of income,
E = cost of education: Ed is the sum of the costs of receiving a degree and
End is the sum of the costs of classes, without receiving a degree,
A = ability,
H = experience, H2 = experience squared,
S = sex (1 = male),
R = race (1 = Caucasian/Asian),
and undefinable random differences or residuals (e).

Data Files. The Colorado Commission on Higher Education (CCHE) has
developed a student records data base (SURDS) that contains records of all
students attending public higher education in Colorado since 1986 and a computer
program (The Cohort Program) that can combine information from different
student files and follow students through the system for seven years. Many of the
regression equations variables are available in the SURDS file (i.e., age, race,
degree obtained, credit hours attempted, sex, and Grade Point Averages [GPAs]
from previous schooling). Table 3.1 at the end of this chapter displays the raw
data elements as they are represented in SURDS and how they contribute to the
variables in the regression equation.
Unemployment insurance (UI) files provide quarterly wage information for
each individual working in Colorado who is not a federal employee or a religious
worker. Self-employed individuals are not required, but have the option to
contribute to this fund. According to the Colorado Labor Market Information
agency, very few self-employed individuals participate in unemployment insurance
and therefore, very few self-employed individuals are contained within the file.
The UI files provided income data from fourth quarter 1990 through fourth
quarter 1992 to supply from six to eight quarters of income data received after
attending higher education, depending on which semester (summer, fall or spring)
the individual last attended school.
Independent Variables. The proposed sample consists of students who
have graduated or taken classes from public postsecondary education institutions in

1991, are not present in the data base after 1991 (and therefore are presumed to
have left school and entered the work force), and have all of the variables present
in their SURDS file. There were 291,477 individual students in the 1990/91 file,
of which 110,527 did not appear in the 1991/92 and 1992/93 files. Of the
110,527 students, 86,847 were classified as Colorado residents in 1991. The final
number of records in the sample equaled 71,657, because of missing or unusable
Data screening of the variables resulted in a decrease of approximately
15,000 records. The primary sources of error were a) incorrect birth dates, b) no
resident credits reported, and c) cost datas discipline did not match the discipline
designation in the student credit hour data. For example, Metropolitan State
Colleges nursing program offers only the last two years of a BS program.
Although the cost data reflected the program structure, there were many student
records that labeled nursing students at MSCD as freshmen or sophomores.
The CCHE SURDS data base contains most of the independent variables
used in the study: demographic variables (sex, race), individuals high school
GPAs, and baccalaureate and masters GPAs for graduate students (ability),
degrees received, number of credit hours attempted (education), and age
(experience). Financial records credit hour costs can be used to develop costs
for each discipline by year by institution and costs for governing boards and the
coordinating board.

The education (E) variable will be defined in two ways: the costs to the
state of credit hours attempted of those individuals that attained a degree Ed and
the costs to the state of credit hours attempted of those individuals that simply
took classes, but did not earn a degree E^. Comparing the returns on these two
variables will indicate if (1) there is evidence for the signaling theory that
receiving a degree signals employers of superior ability (Mare & Winship cited in
Heckman & Singer, 1985) -- but the actual taking of courses is not rewarded in
the market, or (2) as has been found by some recent research (Kane & Rouse,
1992), that a certain number of classes (18-24 credit hours) is remunerative. The
amount of classes taken, without receiving a degree, was divided into three
categories: (1) less than seventeen credit hours, (2) 18-24 credit hours, (3) 25 or
more credit hours at both the undergraduate and graduate level.
Costs for the two education variables were calculated in the same way.
Budget information provides the average state cost per student for each year at
each institution. Each institution also provides information on the amount of
faculty time spent with students (student/faculty ratios) in each discipline at each
level (lower level freshman and sophomore, upper-level juniors and seniors,
graduate 1 masters, and graduate 2 doctoral and first professional students.) In
higher education, salaries constitute 60-75 % of typical budgets and the most
critical salaries in the education production function are faculty salaries. The
amount of faculty time spent educating students determines the majority of faculty

teaching costs and often determines other costs, like support staff salaries, library
collection costs, and supplies.
The cost of a faculty member teaching a three-hour, 300 student freshman
class will be relatively small allocated among all the students enrolled. The same
faculty members salary distributed among 10 students in a three-hour, doctoral
seminar will generate much higher costs per student. For this study, the following
algorithm calculated costs by level and discipline. Total costs distributed by
weighting them by student/faculty ratios by level and discipline gives a range of
costs per credit hour (e.g., if the student/faculty ratio is 10 to 1 for discipline A
and disciplines B ratio is 20 to 1, it requires twice as many faculty salary dollars
per student to teach discipline A).
State operating costs also include indirect costs spent on administration at
the governing board and state level. The cost variable includes costs per credit
hour for governing boards for each of the governed institutions, as well as
CCHEs costs per credit hour for all students in the public system.
The file containing student information contains the institution attended,
fiscal year, student credit hours attempted, and discipline. The costs per
discipline per year per institution per credit hour were then multiplied by the
number of credit hours attempted. In fiscal years 1990 and 1991, both fall and
spring credit hours attempted are included in the file, but for the three previous
fiscal years, only fall credit hours are included and the fall credit hours were
doubled to approximate total annual credit hours.

The ability variables are defined as the GPAs of the former level of
education. For baccalaureate and a few certificate and associate students, the high
school GPA is available or the transfer GPA from a previously-attended
postsecondary institution. For masters or first professional students, the
baccalaureate GPA is available, while the PhD students have a masters (or
occasionally, a first professional) GPA in the file. Although the ability to succeed
at college-level classes requires many attributes persistence, long-term view,
familial or cohort support -- the most easily measured and available variable (for
this and many former studies) to approximate a persons ability to succeed in
college is the former level GPA. Most 2-year students do not have a GPA
variable recorded in the file because community colleges have an open admissions
policy as directed by statute. Therefore, two different files were analyzed: a file
that had the ability variable and one that did not. Comparisons between the two
files provide a measure of the contribution to income supplied by ability, as well
as expanding the research to include all public institutions students.
Mincer (1974) held that experience instead of age was a critical variable
that significantly affected the estimated coefficients of the independent variables.
Without individual data, many economists use an age minus schooling equals
experience equation. Although this algorithm may be appropriate for traditional-
aged students, it often understates the experience of non-traditional students. This
research therefore modified its experience variable algorithm based on the
increased number of non-traditional students in the higher education population

who continue working. The algorithm for this study is: age minus schooling if the
schooling were full-time or if individuals made less than minimum wage for time
periods where income data were available. This more accurately reflects the
change in students since the 1960s when the majority of students participating in
higher education were in their teens and early twenties and attended college on a
full-time basis.
Today, there are many students who attend school, but also work full-time
and thus accrue another year of experience simultaneously. Working adults attend
higher education in increasingly larger numbers every year These individuals
experience variable is understated if this research used the traditional algorithm of
age minus schooling.
An algorithm tested if students were attending full-time (30 credit hours
annually for baccalaureate, associate, or certificate individuals or 24 credit hours
for graduate work). A second test examined the income received in the final two
years of schooling in 1990 or 1991. If the income received was in excess of an
annual salary equal to minimum wage, then the assumption was made that the
person gathered another year of experience. As a result, experience values range
from 0 to 66 years.
Formerly, ethnicity had a significant negative effect on income when
compared to white males incomes (Becker, 1964; Nord, 1986). The effect is not
as straightforward now. Hispanic, Black, and American Indian individuals
historically earned less income than white males because of a lack of education,